Option to impute SNPs

[faithokamoto]

This PR adds a --impute option which will impute all substitutions to ambiguous bases, given that the parental state is consistent with the ambiguity. (For example, N may be any non-gap character, but Y may only be C or T.) It erases these mutations from the relevant nodes. Having the ambiguity never be introduced is the most parsimonious scenario. If a NucMut consists of ambiguous and non-ambiguous bases, the non-ambiguous bases are retained, and merged together if consecutive.

Besides creating impute.cpp and adding the necessary lines to panman.hpp, I also introduce various small conveniences. Examples:

• NucMuts now have functions such as .getNucCode(int i) to do the bitshift lookup of the base at a given position, or .length() to do the bitshift lookup of the length of the mutation. I changed code elsewhere to use these new helper functions, which improves readability and follows the principle of DRY.
• I introduce a hashable struct Coordinate which stores just a position and may be made by copying off a NucMut. I use this struct to speed up consolidateNucMutations() by converting it to use an unordered_map.
• I factor out functions to apply and undo nucleotide/block mutations given a set of mutations and a tracker (e.g. a sequence_t). These were used in two separate FASTA-printing functions, and I use them for my imputation function.

I also perform one unambiguous bugfix, in mergeNodes(), which now e.g. updates the parent attribute of all new children.

[faithokamoto]

You can ignore the install/installationUbuntu.sh and workflows/scripts/wfmash.sh updates, I'm honestly not quite sure how they're different. I reverted the -mavx512f flag in CMakeLists.txt because it caused segfaults on my machine (Windows 11 Home running Ubuntu in a WSL) due to a bad instruction set.

[faithokamoto]

Ahhh I triple-checked and I must've broken something. The imputation isn't working like it was yesterday.

[faithokamoto]

nevermind it is all good