Bug Fix: cut_point not in exon_positions

[mbowcut2]

Bug Summary

When a sgRNA cut point falls before the start of the exon (cut_point < exon_positions[0]), the amino acid cut point
calculation produces a negative value:

  amino_acid_cut_point = (cut_point - refs[ref_name]['exon_positions'][0] + 1) // 3                                          
  # e.g., (97 - 100 + 1) // 3 = -2 // 3 = -1, or worse cases → -3                                                            

This negative value is then passed to get_amino_acid_row() which does row['ref_positions'].index(-3), and since
ref_positions never contains negative values, it throws ValueError: -3 is not in list.

Changes

1. CRISPResso2/CRISPRessoCORE.py line 6121 (root cause fix): Clamp amino_acid_cut_point to valid range [0,
   len(coding_seq_amino_acids) - 1]. This handles both the case where the cut is before the exon (clamp to 0) and after the
   exon (clamp to last amino acid).
2. CRISPResso2/CRISPRessoShared.py line 1540 (defensive fallback): Wrap row['ref_positions'].index(amino_acid_cut_point) in
   a try/except ValueError that falls back to cut_idx = 0, so any other code path that passes an out-of-range value won't crash either.

[mbowcut2]

another round of changes:
Summary of Changes

CRISPResso2/CRISPRessoCORE.py (Steps 1, 2, 3)

Problem: The amino acid table code (Figure 9a) was outside the sgRNA loop but used stale cut_point, plot_cut_point, and new_sgRNA_intervals variables from the last sgRNA iteration.
When the last sgRNA's cut point was far from the coding region, the amino_acid_cut_point was wildly out of range.

Fix:

1. Best sgRNA selection — For each coding sequence, finds the sgRNA whose cut point is closest to that coding sequence's exon interval (distance=0 if inside, otherwise distance to
   nearest edge). Falls back to exon midpoint if no guides exist.
2. Per-coding-seq exon start — Changed the formula from (cut_point - refs[ref_name]['exon_positions'][0] + 1) // 3 to (best_cut_point - exon_start + 1) // 3, using the specific exon
   interval start rather than the merged exon union start.
3. Removed unused params — Removed sgRNA_intervals, sgRNA_names, sgRNA_mismatches from the plot dict (they were passed through but never used by plot_amino_acid_heatmap). Used
   best_plot_cut_point instead of the stale plot_cut_point.

CRISPResso2/CRISPRessoShared.py (Step 4)

Problem: The except ValueError: cut_idx = 0 fallback was crude — jumping to position 0 when the cut point position is deleted in an allele.

Fix: When amino_acid_cut_point isn't found in ref_positions, find the closest valid position instead of defaulting to 0. Also narrowed the bare except: on gap_incentive[cut_idx] to
except IndexError:.

CRISPResso2_tests expected results (Step 5)

Updated 4 amino acid table files + 2 HTML reports. The amino acid tables revert from the HEAD~1 interim values (which used amino_acid_cut_point=16 from the clamp-to-last-sgRNA
approach) back to values matching the correct sgRNA selection (amino_acid_cut_point=12, using the primary guide that's inside the exon). The HTML changes add new parameters
(amplicon_coordinates, vcf_output) from the VCF output feature.

Verification (Step 6)

Both make params test and make base_editor test pass cleanly.

[Colelyman]

I think this PR is on the right track, but needs some refinement.

(And quick clarification in the description, ref_positions does have negative values, but it is when there is an insertion that isn't represented in the original sequence. Not super relevant to the content or design of the PR, this still needs to be fixed.)