fix: fail with a GTF-specific error when no genes can be extracted

[BenjaminDEMAILLE]

Fixes #126.

Problem

When a GTF yields zero genes, parse_gtf only logged a WARN and returned an empty map. The run then died much later in count_reads with the chromosome-mismatch diagnosis, which cannot apply — there is nothing on the GTF side to mismatch against:

    Parsing GTF annotation...
[WARN  rustqc::gtf] No genes extracted from GTF file './truncated.gtf'. ...

  ✗ error: Failed to process sample.bam: Chromosome name mismatch: no reads could be assigned to any gene.

Alignment chromosomes (first 5): chr1, chr10, chr11, chr12, chr13
GTF chromosomes (first 5):

...
chromosome_mapping:

The suggested fix (chromosome_mapping) is empty and cannot help, and the actual cause is a WARN line that is easy to miss.

Fix

parse_gtf now records why nothing was extracted while scanning, and returns a fatal error naming the most likely cause:

Situation	Message
No non-comment lines	the file contains no annotation lines (it is empty, or every line is a comment)
Every line has < 9 columns	all N annotation lines have fewer than the 9 mandatory tab-separated GTF columns (truncated file, or space- instead of tab-separated)
No exon features	no 'exon' features were found; feature types present: gene, mRNA. GFF3 files ... are not supported — convert to GTF
Exons but no gene_id	N feature lines were found but none carried a 'gene_id' attribute

count_reads keeps a guard for library callers: an empty annotation index now produces an empty-annotation error rather than the chromosome-mismatch text.

Behaviour change

An empty annotation is now a hard error at parse time instead of a warning. This is the same outcome as before (the run always failed a few seconds later), just reported at the point where the cause is known.

Verification

Reproducing the issue's steps with cut -f1-5 tests/data/test.gtf:

    Parsing GTF annotation...
Error: No genes could be extracted from GTF file '.../trunc.gtf': all 22 annotation lines have
fewer than the 9 mandatory tab-separated GTF columns (truncated file, or space- instead of
tab-separated).

Every RustQC analysis needs a usable gene annotation, so processing cannot continue.
Check that the file is uncorrupted, tab-separated GTF (not GFF3) with 'exon' features carrying
'gene_id' attributes.

Four new unit tests cover the truncated, GFF3, missing-gene_id and empty-file cases.

• cargo test — 204 lib + 18 integration tests pass
• cargo fmt --check, cargo clippy -- -D warnings clean

🤖 Generated with Claude Code