Reassign alleles and Genotype Inference improvement

[ayeletperes]

Speeds up allele reassignment and moved genotype inference functions from enchantr

Changes

reassignAlleles

1. Now uses the alakazam Rcpp 'hamming' distance function. The call is guarded by a
   runtime exists(..., envir = asNamespace("alakazam")) check and verified to
   reproduce the old result bit-for-bit, so it degrades gracefully to pure R if
   alakazam hasn't been updated yet.
2. New flags strip_d and reassign_uncalled. These preserve the old function
   behavior, where it annotates empty cells and strips the D of duplicated genes
   (e.g. IGHV1-69D).

genotypeFasta

1. New flag strip_d to avoid stripping the D of duplicated genes. Default preserves
   old function behavior.
2. New flag include_unseen to include germline DB alleles for genes not present in
   the genotype. Default preserves old function behavior.

inferGenotypeBayesian

1. New flag genotyped_alleles computes the alleles that passed the genotype and adds
   a new genotyped_alleles column (moved from enchantr).
2. Added support for mixed (multi-chain) data, computing depth based on the locus
   column.

plotGenotypeConfidence

1. New plot visualizing the k_diff column of the Bayesian inference, based on the old
   plot function.

plotGenotype

1. New flag allele_col to plot the genotyped_alleles column from the Bayesian
   inference; previously hard-coded to alleles.

Notes

Please approve after the alakazam PR (add-mutation-count-rcpp) is approved. reassignAlleles relies on the Rcpp hamming-distance functions exported there for its speedup. It is functionally non-breaking without them (falls back to pure R), but the intended path needs alakazam in first.