feat: E2E benchmark halo2 generate flamegraphs

[stephenh-axiom-xyz]

Implements the following:

• bench-metrics feature properly propagates to openvm-native-compiler module when set in the openvm-benchmarks and/or openvm-sdk modules.
• SDK supports profiling for halo2 proof generation.
• Flamegraph generation script a) generates halo2 flamegraphs and b) doesn't attempt to generate flamegraphs if the benchmark doesn't exist

Latest fib_e2e run results here (note fixed_cells is zero throughout and thus doesn't show up).

[jonathanpwang]

Can we make a single flamegraph to consolidate lookup_advice_cells + simple_advice_cells? they should both be counted towards total cell count

[jonathanpwang]

LGTM after addressing comments

[github-actions]

group	app.proof_time_ms	app.cycles	app.cells_used	leaf.proof_time_ms	leaf.cycles	leaf.cells_used
verify_fibair	(+302 [+8.7%]) 3,777	(+34974 [+4.9%]) 746,301	(+4500079 [+17.6%]) 30,011,454	-	-	-
fibonacci_program	(-19 [-0.3%]) 6,037	1,500,137	51,505,102	-	-	-
regex_program	(-368 [-1.9%]) 18,563	4,190,904	165,028,173	-	-	-
ecrecover_program	(+15 [+0.6%]) 2,617	285,401	15,092,297	-	-	-

Commit: ada40bd

Benchmark Workflow