Speed-up `bench-gas` check

[danceratopz]

Currently, the just bench-gas check is the bottleneck in CI at 18 minutes.

One possible solution is to reduce the --gas-benchmark-values flag from 1.0 to 0.5 (or even 0.2)? Perhaps this causes problems with tests, we should try it.

Currently, changing this results in this pydantic validation error:

INTERNALERROR> pydantic_core._pydantic_core.ValidationError: 1 validation error for GasBenchmarkValues
INTERNALERROR> 0
INTERNALERROR>   Input should be a valid integer, unable to parse string as an integer [type=int_parsing, input_value='0.5', input_type=str]
INTERNALERROR>     For further information visit https://errors.pydantic.dev/2.12/v/int_parsing

which can fix by changing the type from int to float > 0.001 (or something else more reasonable) - this flag is measured in Mgas.

[shrirajpawar4]

@danceratopz I investigated this and I’d like to work on it.

The issue appears to be in the benchmark CLI/parsing layer rather than in the benchmark tests themselves. --gas-benchmark-values is documented in Mgas, but GasBenchmarkValues currently validates
List[int], so values like 0.5 fail before benchmark generation starts.

My proposed fix is:

1. Accept decimal Mgas inputs for --gas-benchmark-values.
2. Parse them with Decimal and convert them to integer gas units before injecting gas_benchmark_value.
3. Keep downstream benchmark code integer-based, since the tests and gas validation logic expect an absolute gas budget in gas units.
4. Update benchmark IDs / output subdirectory formatting so fractional values are represented cleanly.
5. Change just bench-gas from --gas-benchmark-values 1 to 0.5.
6. Add tests for decimal inputs and update the benchmark docs.

If this is not already in progress, I can open a PR for it.

[shrirajpawar4]

@danceratopz opened #2694

[shrirajpawar4]

@LouisTsai-Csie would appreciate if you could take a look at this as well

[danceratopz]

Thanks a lot for trying this @shrirajpawar4, please check my review on your PR, #2694 (review)