gtfsdb: compute batch size via SQLite limit and fail fast on invalid …

[Adityatorgal17]

Summary

Follow-up to #517 addressing review feedback.

Changes

• Remove BulkInsertBatchSize field and DefaultBulkInsertBatchSize constant
• Compute batch size via SafeBatchSize using SQLite variable limit (32766)
• Add stopTimeFieldsPerRow=10 and shapeFieldsPerRow=5 constants for clarity
• Panic on fieldsPerRow <= 0 (programmer error)
• Update stale comments in bulkInsertShapes
• Update TestSafeBatchSize to assert panic for invalid input

Notes

fieldsPerRow is a compile-time constant at call sites, so invalid values
represent programmer error and should fail fast.

Follow-up to #517.

[aaronbrethorst]

Hey Aditya, this is a really clean follow-up. Removing the dead BulkInsertBatchSize/DefaultBulkInsertBatchSize code, switching to a panic for what can only ever be a programmer error, and extracting the magic numbers into named constants — each change makes the code clearer and more honest about its invariants. I verified that stopTimeFieldsPerRow = 10 and shapeFieldsPerRow = 5 both match their respective INSERT column lists, placeholder counts, and args-per-row exactly. The updated batch size comment in bulkInsertShapes is also accurate (50 * 6553 = ~327k).

There's nothing left to change — this PR is ready to merge. Nice work keeping the follow-up focused and tidy.