feat(cuda): VPMM v3 and async app prover

[gaxiom]

No description provided.

[jonathanpwang]

I merged #2153 into this branch; we can close the latter after testing is done

[Copilot]

Pull Request Overview

This PR adds VPMM v3 support and implements an asynchronous app prover for improved concurrent proof generation. The changes focus on enabling async proof generation capabilities, adding concurrency controls for execution suspension, and updating dependencies.

• Introduces AsyncAppProver for thread-safe concurrent proof generation
• Adds execution suspension logic via should_suspend context checks
• Updates stark backend dependencies to v1.2.1-rc.6 and adds async test infrastructure

Reviewed Changes

Copilot reviewed 9 out of 11 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
crates/sdk/src/prover/app.rs	Implements AsyncAppProver with semaphore-based concurrency control
crates/vm/src/arch/interpreter.rs	Adds suspension logic to TCO execution path
extensions/keccak256/circuit/src/tests.rs	Adds multi-threaded async test for CUDA tracegen
Cargo.toml	Updates stark backend dependencies to rc.6 and adds tokio workspace dependency
crates/sdk/Cargo.toml	Adds async feature with tokio dependency
benchmarks/prove/Cargo.toml	Updates clap/tokio to workspace versions and adds async feature
.github/workflows/	Updates CI configuration for async testing and Foundry version

---

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

[github-actions]

group	app.proof_time_ms	app.cycles	app.cells_used	leaf.proof_time_ms	leaf.cycles	leaf.cells_used
verify_fibair	(+35 [+15.6%]) 259	322,610	2,058,654	-	-	-
fibonacci	(-6 [-0.6%]) 1,007	1,500,209	2,100,402	-	-	-
regex	(+5 [+0.2%]) 2,328	4,137,502	17,695,216	-	-	-
ecrecover	(-8 [-1.1%]) 703	122,859	2,262,718	-	-	-
pairing	(+9 [+0.6%]) 1,453	1,745,742	25,468,210	-	-	-

Commit: 0ca84f8

Benchmark Workflow