De-jankify implementation of DOP in polymer building

[timbernat]

The current way DOP (degree of polymerization) is defined in polymers.building.build_linear_polymer() is highly non-intuitive and case-specific to an AB/BA (i.e. diblock) sequenced linear polymer. Planned Improvements to this include:

• Remove the DOP - 2 hardcode
• Replace DOP input with num_monomers input (which is much less ambiguous)
  • Correcting internals with explicit calculation based on requested DOP, sequence length, and terminal groups, namely DOP_mbuild = (num_monomers - num_terminal) / len(sequence), with failure when this is not an int (i.e. incompatible sequence)
  • Provide mechanism to count how many term groups are provided in a MonomerGroup
  • Correct logging of target DOP to reflect these changes
  • Provide mechanicm for fractional sequence repeats (i.e. 2.5 * "BACA" -> "BACA|BACA|BA")
• Improve accuracy of .estimation #atom calculations
  • Don't calculate over all monomers present in a MonomerGroup, only those used for actually building
• Deprecate "DOP" keyword in favor of more explicit "n_monomers"
  • Generally, tighten up terminology such as "n_monomers", "DOP", "chain length", etc throughout .building and .estimation
• Clean up behavoir of "term_orient" attribute of MonomerGroup
  • Invert order of fields, such that the keys are from {'head', 'tail'}, and the values are the names of monomers
  • Add property to generate end group RDMols based on the value of term_orient
  • Add auto-listification of single-SMARTS entries to monomers on init to enforce stupid legacy-inherited listification
• Write units tests for building and estimation

[timbernat]

Addressed by #34