Introduce _cheminfo_components_smiles

[merkys]

This PR introduces _cheminfo_components_smiles alongside _cheminfo_components_stdinchi. At the same time it renames _cheminfo_components_stdinchi_counts to _cheminfo_components_counts to be reused for both _cheminfo_components_smiles alongside _cheminfo_components_stdinchi. Thus now there will be a possibility to express components both using standard InChI and SMILES.

[vaitkus]

Not sure _cheminfo_components_stdinchi_counts should be changed into _cheminfo_components_counts. Depending on what SMILES canonicalisations are used, SMILES may consider two molecular entities as different where standard InChI does not (e.g. standard InChI does not discern between tautomers).

[a-e-day]

I was thinking that it might be good to keep them separate (there may be some ambiguity in how people define the count and they might work out inchis and smiles separately)

[merkys]

After today's workshop discussion I am requalifying this PR as a draft. I have summarized the alternative proposals which were discussed today in #17.