OpenMM 8.4 validation

[jthorton]

Issue to track the results of various benchmarks we are running to validate OpenMM 8.4.

Notes

• All benchmarks now use fast settings by default
• Inputs and results can be found at openfe-benchmarks
• Lomap industry benchmark networks are used
• Sage-2.3.0 small molecule forcefield is used with the AshGC charge model for all 8.4 runs and 8.3 Jacs runs
• Charge change results with OpenMM 8.1.1 are taken from the industry benchmarks paper and use ambertools charges and Sage-2.2.0 with the old default simulation settings (not fast settings).

[jthorton]

TYK2 Jacs set hybrid topology RBFE OpenMM 8.4 vs 8.3

P38 Jacs set hybrid topology RBFE OpenMM 8.4 vs 8.3

Irak4_s3 charge change edges OpenMM 8.4 vs 8.1.1 (Industry benchmarks results)

Irak4_s2 charge change edges OpenMM 8.4 vs 8.1.1 (industry benchmark results)

egfr charge change edges OpenMM 8.4 vs 8.1.1 (industry benchmark results)