[Feature, Performance] Optimize HamEvo for commuting Hamiltonians

[jpmoutinho]

Moving the original issue from Qadence: pasqal-io/qadence#346

With the reorganization of HamEvo, this would make sense to implement in PyQTorch instead.

This is essentially equivalent to doing exact Trotterization automatically whenever possible. Should be properly planned in that context as well, since we have Trotterization code in Qadence.