Faster MCS Calculation

[kexul]

The current matchAtoms() uses the entire molecule to get MCS, a simple idea to accelerate it is to get MCS for heavy atoms and then match up hydrogens hanging off the MCS. See this implementation for example: https://github.com/OpenFreeEnergy/Lomap/blob/84e1bb0d20ceeed835166fe7218dd21f04248030/lomap/mcs.py#L1170

I didn't quantitatively test the speedup(no benchmark found) but it performed very well for large molecules in my project ( the current matchAtoms() often times out).