[Crum] Design a Plan for Override Types

[pishoyg]

There are two purposes:

1. There is some loss of information in Marcion's representation, such as:
   • Dual-gender nouns that assume one gender more often than the other, or assumed one of the two genders only once. (We had a poor attempt at implementing that by introducing a female / male type, as opposed to the current male / female type, but this isn't good enough.)
2. More information is needed to enable a more rigorous inflection module, such as:
   • Whether a noun accepts a definite / indefinite article (a word such as ⲁⲩⲣⲏϫ//, for example, doesn't).

This data is to be treated as an appendix, not typos (see #137). They shouldn't override Marcion's data, but should augment it.

[pishoyg]

126 concerns manual data collection, this concerns the construction of the pipeline that enables it.

We already have a column in the Crum Appendices sheet and the local TSV mirror. But the column is empty, unused, and doesn't have a corresponding statistic in stats.sh.

[pishoyg]

One could argue that, for the use case that is dual-gender nouns, it's actually beneficial for users to be able to see that in the interface, not just for the sake of rigor.