fix: case-sensitive json marshalling and unmarshalling

[nacx]

Description

Use strict json marshalling and unmarshalling configuration.

Related Issues/PRs (if applicable)

N/A

Special notes for reviewers (if applicable)

N/A

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 84.38%. Comparing base (a833780) to head (d81791e).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1863   +/-   ##
=======================================
  Coverage   84.38%   84.38%           
=======================================
  Files         130      130           
  Lines       18114    18114           
=======================================
  Hits        15285    15285           
  Misses       1883     1883           
  Partials      946      946           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[yuzisun]

Can we make this configurable? it may break requests that are passing unknown elements or fields not defined on struct

[nacx]

I've updated the change to focus on case-sensitivity, which shouldn't affect unknown fields, etc, that we rely on.

[nacx]

/retest

[nacx]

@yuzisun can you take a look and approve? This is passing all tests and fixes the security advisory on MCP.

[mathetake]

needs benchmark comparison just in case

https://github.com/envoyproxy/ai-gateway/blob/45ae2fef8647e9c0d985bce43a7f022c8499493e/tests/data-plane/bench_test.go#L28-L34

[nacx]

❯ benchstat old.txt new.txt
goos: darwin
goarch: arm64
pkg: github.com/envoyproxy/ai-gateway/tests/data-plane
cpu: Apple M4 Pro
                                                                                        │    old.txt    │               new.txt               │
                                                                                        │    sec/op     │   sec/op     vs base                │
ChatCompletions/openai/10-messages/10-byte-per-message/small-response-12                  105.27µ ±  3%   81.54µ ± 1%  -22.54% (p=0.000 n=10)
ChatCompletions/openai/10-messages/1000-bytes-per-message/mediaum-response-12              261.2µ ±  1%   215.2µ ± 2%  -17.62% (p=0.000 n=10)
ChatCompletions/openai/10-messages/100000-bytes-per-message/large-response-12              2.221m ±  2%   1.881m ± 2%  -15.29% (p=0.000 n=10)
ChatCompletions/openai/10-messages/1000000-bytes-per-message/large-response-12             6.261m ±  1%   5.469m ± 2%  -12.65% (p=0.000 n=10)
ChatCompletions/openai/1000-messages/1-byte-per-message/small-response-12                  219.2µ ±  1%   164.3µ ± 2%  -25.06% (p=0.000 n=10)
ChatCompletions/openai/1000-messages/100-bytes-per-message/mediaum-response-12             408.1µ ±  2%   317.9µ ± 1%  -22.08% (p=0.000 n=10)
ChatCompletions/openai/1000-messages/10000-bytes-per-message/large-response-12             6.576m ±  3%   5.658m ± 3%  -13.96% (p=0.000 n=10)
ChatCompletions/openai/100000-messages/1-byte-per-message/large-response-12                12.56m ±  2%   11.09m ± 4%  -11.75% (p=0.000 n=10)
ChatCompletions/openai/500000-messages/1-byte-per-message/large-response-12                68.15m ±  7%   57.27m ± 8%  -15.97% (p=0.000 n=10)
ChatCompletions/aws-bedrock/10-messages/10-byte-per-message/small-response-12             100.66µ ±  2%   77.06µ ± 1%  -23.44% (p=0.000 n=10)
ChatCompletions/aws-bedrock/10-messages/1000-bytes-per-message/mediaum-response-12         118.3µ ± 16%   103.9µ ± 2%  -12.19% (p=0.029 n=10)
ChatCompletions/aws-bedrock/10-messages/100000-bytes-per-message/large-response-12         735.8µ ±  1%   721.4µ ± 1%   -1.95% (p=0.000 n=10)
ChatCompletions/aws-bedrock/10-messages/1000000-bytes-per-message/large-response-12        4.660m ±  1%   4.617m ± 1%        ~ (p=0.089 n=10)
ChatCompletions/aws-bedrock/1000-messages/1-byte-per-message/small-response-12             213.8µ ±  1%   210.7µ ± 1%   -1.44% (p=0.001 n=10)
ChatCompletions/aws-bedrock/1000-messages/100-bytes-per-message/mediaum-response-12        286.1µ ±  1%   286.3µ ± 1%        ~ (p=0.912 n=10)
ChatCompletions/aws-bedrock/1000-messages/10000-bytes-per-message/large-response-12        5.086m ±  4%   5.030m ± 4%        ~ (p=0.631 n=10)
ChatCompletions/aws-bedrock/100000-messages/1-byte-per-message/large-response-12           14.32m ±  2%   14.62m ± 4%        ~ (p=0.247 n=10)
ChatCompletions/aws-bedrock/500000-messages/1-byte-per-message/large-response-12           95.74m ±  8%   95.49m ± 6%        ~ (p=0.853 n=10)
ChatCompletions/gcp-vertexai/10-messages/10-byte-per-message/small-response-12             79.42µ ±  2%   78.09µ ± 2%        ~ (p=0.075 n=10)
ChatCompletions/gcp-vertexai/10-messages/1000-bytes-per-message/mediaum-response-12        168.2µ ±  1%   167.9µ ± 2%        ~ (p=0.436 n=10)
ChatCompletions/gcp-vertexai/10-messages/100000-bytes-per-message/large-response-12        1.167m ±  1%   1.165m ± 0%        ~ (p=0.971 n=10)
ChatCompletions/gcp-vertexai/10-messages/1000000-bytes-per-message/large-response-12       4.975m ±  2%   4.983m ± 1%        ~ (p=0.631 n=10)
ChatCompletions/gcp-vertexai/1000-messages/1-byte-per-message/small-response-12            199.7µ ±  1%   195.9µ ± 1%   -1.92% (p=0.000 n=10)
ChatCompletions/gcp-vertexai/1000-messages/100-bytes-per-message/mediaum-response-12       337.4µ ±  1%   330.6µ ± 1%   -2.01% (p=0.000 n=10)
ChatCompletions/gcp-vertexai/1000-messages/10000-bytes-per-message/large-response-12       5.440m ±  1%   5.418m ± 2%        ~ (p=0.481 n=10)
ChatCompletions/gcp-vertexai/100000-messages/1-byte-per-message/large-response-12          13.16m ±  5%   12.98m ± 4%        ~ (p=0.436 n=10)
ChatCompletions/gcp-vertexai/500000-messages/1-byte-per-message/large-response-12          82.56m ±  9%   86.13m ± 5%   +4.33% (p=0.035 n=10)
ChatCompletions/gcp-anthropicai/10-messages/10-byte-per-message/small-response-12          85.27µ ±  2%   84.16µ ± 2%        ~ (p=0.247 n=10)
ChatCompletions/gcp-anthropicai/10-messages/1000-bytes-per-message/mediaum-response-12     148.8µ ±  1%   145.7µ ± 1%   -2.06% (p=0.000 n=10)
ChatCompletions/gcp-anthropicai/10-messages/100000-bytes-per-message/large-response-12     1.807m ±  1%   1.787m ± 1%   -1.10% (p=0.002 n=10)
ChatCompletions/gcp-anthropicai/10-messages/1000000-bytes-per-message/large-response-12    14.11m ±  2%   13.87m ± 2%        ~ (p=0.063 n=10)
ChatCompletions/gcp-anthropicai/1000-messages/1-byte-per-message/small-response-12         418.6µ ±  2%   406.1µ ± 0%   -2.98% (p=0.000 n=10)
ChatCompletions/gcp-anthropicai/1000-messages/100-bytes-per-message/mediaum-response-12    628.9µ ±  0%   622.6µ ± 1%   -1.00% (p=0.002 n=10)
ChatCompletions/gcp-anthropicai/1000-messages/10000-bytes-per-message/large-response-12    14.50m ±  3%   14.28m ± 2%        ~ (p=0.247 n=10)
ChatCompletions/gcp-anthropicai/100000-messages/1-byte-per-message/large-response-12       43.22m ±  3%   42.18m ± 3%        ~ (p=0.165 n=10)
ChatCompletions/gcp-anthropicai/500000-messages/1-byte-per-message/large-response-12        1.071 ±  2%    1.073 ± 1%        ~ (p=0.912 n=10)
ChatCompletionsStreaming/OpenAI_Streaming-12                                               20.02m ±  0%   20.21m ± 1%   +0.98% (p=0.002 n=10)
ChatCompletionsStreaming/AWS_Streaming-12                                                  20.01m ±  2%   20.20m ± 1%   +0.96% (p=0.000 n=10)
geomean                                                                                    2.117m         1.991m        -5.94%

[mathetake]

Nice

[nacx]

@mathetake can you stamp? :)

[nacx]

/retest