perf: Improve GKR memory usage

[Tabaie]

This PR introduces a simple memory pool to massively reduce the number of heap allocations of single field elements introduced by the new GKR API.

On the benchmark std/hash/mimc/gkr-mimc/BenchmarkMerkleTree and a similar one for the gkr-mimc currently on the Linea monorepo, with an hpc6a.48xlarge instance, we get:

This PR: 3937 MB
feat/gkr/hashes: 20292 MB (5.15x improvement)
linea-monorepo@main: 7229 MB (1.83x improvement)

[Copilot]

Pull Request Overview

This PR improves memory usage in GKR implementation by introducing a memory pool for field elements to reduce heap allocations. The optimization achieves significant memory reduction: from 20292 MB to 4403 MB (4.6x improvement) on benchmarks.

Key changes include:

• Implementation of a synchronous memory pool system in gateAPI to reuse field element allocations
• Replacement of global API instances with local instances across all curve implementations
• Simplification of test benchmarks by removing complex file caching logic

Reviewed Changes

Copilot reviewed 28 out of 28 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
std/hash/mimc/gkr-mimc/gkr-mimc_test.go	Renamed circuit types and simplified benchmark by removing file caching
std/permutation/gkr-mimc/gkr-mimc_test.go	Renamed circuit types for consistency
internal/gkr/*/gkr.go	Added memory pool to gateAPI with element allocation/deallocation methods
internal/gkr/*/gate_testing.go	Updated to use local gateAPI instances
internal/gkr/*/solver_hints.go	Added local gateAPI instances with proper cleanup
internal/generator/backend/template/gkr/*.tmpl	Updated templates to generate optimized code

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[Tabaie]

Closed in favor of #1616