Optimize FastxZipped.__next__ to use a single pass over records

[msto]

Summary

FastxZipped.__next__ currently makes multiple passes over records on every call:

1. all(record is None for record in records) — check if all exhausted
2. all_not_none(records) — check if any are None (truncation)
3. [record.name for record in records] — extract names
4. [self._name_minus_ordinal(name) for name in ...] — strip ordinals
5. set(record_names) — check uniqueness

Since this runs on every record, there may be a performance concern for large FASTX files with many read groups.

These could be consolidated into a single loop that:

• Detects None records (truncation or exhaustion)
• Extracts and validates names in one pass
• Short-circuits on name mismatch

Context

Raised in #259 (review comment by @nh13): #259 (comment)

Notes

The number of elements in records is bounded by the number of FASTX files being zipped (typically 2–4), so the constant factor may be negligible in practice. Worth benchmarking before optimizing.