test(push): pin tabular schema/type-inference boundaries

[LukasWodka]

Summary

First follow-up from the mutation-testing pass. Gremlins found the internal/push SQL type-inference / schema helpers had unguarded boundaries. The schema drives the ingestor's column dtypes, so an off-by-one here silently mis-types a column (e.g. an INT that should be BIGINT overflows on write). These are pure functions, so the fixes are targeted, fixture-free unit tests — each chosen to kill a specific surviving mutant.

Helper	Mutant killed	Test pins
integerType	v > int32Max → >=, v < int32Min → <=	exactly int32Max/int32Min stays INT; one beyond widens to BIGINT
sqlBaseType	i >= 0 → i > 0	a "(255)"-only token has base "" (no-leading-identifier contract)
splitSchemaEntries	depth > 0 → depth >= 0	a stray ) must not drive paren depth negative (else it swallows the next separator)
containsASCIIDigit	>= '0' → > '0', <= '9' → < '9'	'0'/'9' count as digits; '/'/':' (just outside) don't
isIntegerType	== → != (both terms)	INT/BIGINT true, everything else false

All 5 are mutation-proven: flipping the operator in the source makes the matching test fail. Pure additive test file — no production change.

(Equivalent/impractical mutants in the same cluster — e.g. cleanTokens's cap slice and varcharOf's > n where the boundary value produces identical output — were deliberately not chased.)

Test plan

• make ci green (build, vet, gofmt -s, go test -race, errcheck/ineffassign/misspell, schema-check).

🤖 Generated with Claude Code

---

Note

Low Risk
Additive tests only; no runtime or schema logic is modified.

Overview
Adds internal/push/tabular_schema_test.go, a pure test-only follow-up to gremlins mutation testing. It locks down boundary behavior for schema helpers that drive ingestor column dtypes: integerType (INT vs BIGINT at int32 limits and VARCHAR past int64), sqlBaseType (including "(255)" → empty base), splitSchemaEntries (stray ) must not merge entries), containsASCIIDigit (ASCII '0'/'9' edges), and isIntegerType (INT/BIGINT only).

No production code changes — each test is documented as targeting a specific surviving mutant operator flip.

^{Reviewed by Cursor Bugbot for commit cf92070. Bugbot is set up for automated code reviews on this repo. Configure here.}

[LukasWodka]

bugbot run

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit cf92070. Configure here.}

[saadqbal]

LGTM 👍 Genuinely mutation-aware — verified each case kills the mutant it names: the inclusive int32 width boundary, sqlBaseType("(255)")→"" (the i>=0 slice), and the paren-depth floor keeping a),b as two entries are the subtle ones and they all hold. Nice work closing out the audit.