chore(call): truncate max reads instead of skipping locus

docs/caller_usage.md

@@ -8,7 +8,7 @@
 * `--min-reads ##`: Minimum number of supporting reads needed to make a call. **Default:** 4
 * `--min-allele-reads ##`: Minimum number of supporting reads needed to call a specific allele size. 
   **Default:** 2
-* `--max-reads ##`: Maximum number of supporting reads before a locus is skipped. **Default:** 250
+* `--max-reads ##`: Maximum number of supporting reads to use for calling a locus. **Default:** 250
 * `--min-avg-phred ##`: Minimum average PHRED score for relevant bases (flanking region + tandem repeat).
   Read segments with average PHRED scores below this (common with a threshold of ~13 and ONT Ultra Long reads, 
   for example) will be skipped. **Default:** 13

strkit/call/call_locus.py

@@ -948,9 +948,8 @@ def call_locus(
 
     logger_.debug(f"{locus_log_str} - got {n_overlapping_reads} overlapping aligned segments")
 
-    if n_overlapping_reads > params.max_reads:
-        logger_.warning(f"{locus_log_str} - skipping locus; too many overlapping reads")
-        return locus_result
+    if n_overlapping_reads > params.max_reads:  # TODO: sample across full set instead?
+        logger_.warning(f"{locus_log_str} - locus has excess reads, using the first {params.max_reads}; misalignment?")
 
     sorted_read_lengths = np.sort(read_lengths)