chore: add memory report into the CI

[guipublic]

Description

Problem*

Resolves #6604

Summary*

Add a bash script which computes memory peak usage on 3 test programs, and a CI action to include this information on CI.

Additional Context

Documentation*

Check one:

• No documentation needed.
• Documentation included in this PR.
• [For Experimental Features] Documentation to be submitted in a separate PR.

PR Checklist*

• I have tested the changes locally.
• I have formatted the changes with Prettier and/or cargo fmt on default settings.

[github-actions]

Peak Memory Sample

Program	Peak Memory
keccak256	84.02M
workspace	121.85M
regression_4709	348.13M

[guipublic]

In noir-diff, I use a workaround to not download the 'previous artefact', because for the memory report, this part stalls, and I don't know why.
As a result, I do not do a diff with the previous artefact.

[TomAFrench]

because for the memory report, this part stalls, and I don't know why.

This is because there's no report in the master branch to compare against and so it's just looking back through years of git history for one. We could add a bootstrap commit option to the action to not look past a certain commit.

[guipublic]

This is because there's no report in the master branch to compare against and so it's just looking back through years of git history for one. We could add a bootstrap commit option to the action to not look past a certain commit.

What about using a counter and looking back only 100 times?
UPDATE I added the diff ability by adding a column in the report which show the evolution in %, if any (and also I used the '100' counter).

[guipublic]

Peak memory profiling for test programs show that most of the programs are about 80M of peak memory usage, except for the ones below.
For the report I chose:

• Keccak as a small peak, which I think is a good candidate for representing what Noir can be used for
• Workspace for a medium peak, which could indicate frontend issues
• regression_4709 for a higher peak. The test is simple, it is doing a many additions in a nested loop.

Program	Peak Memory
bench_2_to_17	3.1G
regression_5252	3G
ram_blowup_regression	2.8G
eddsa	885M
poseidonsponge_x5_254	396M
regression_4709	387M
hashmap	280M
sha256_var_padding_regression	237M
sha256_regression	190M
regression_4449	185M
sha2_byte	150M
sha256_var_size_regression	130M
workspace	128M
poseidon_bn254_hash_width_3	128M
conditional_1	115M
uhashmap	114M
no_predicates_numeric_generic_poseidon	104M
fold_numeric_generic_poseidon	102M
sha256	93M
keccak256	90M
debug_logs	88M
array_dynamic_blackbox_input	87M
nested_array_dynamic	86M
sha256_var_witness_const_regression	84M