fix(tokenize): use camelCase @SerializedName values on TextAnalyzer + add round-trip test

[g-despot]

Summary

The TextAnalyzer record was annotated with snake_case @SerializedName values (ascii_fold, ascii_fold_ignore, stopword_preset), but Weaviate's REST schema expects camelCase (asciiFold, asciiFoldIgnore, stopwordPreset) — so every per-property analyzer setting was silently dropped on the wire. Verified by comparing the stored schema after Property.text(... textAnalyzer(TextAnalyzer.of(t -> t.foldAscii(true).keepAscii("é")))): pre-patch the property comes back with no textAnalyzer field at all; post-patch it stores { "asciiFold": true, "asciiFoldIgnore": ["é"] }.

This PR flips the three @SerializedName values to camelCase and adds TextAnalyzerITest covering both the schema round-trip (collection.config.get() returns the persisted analyzer) and the behavioral effect (a cafe filter on a folded Café Crème row only matches the property with foldAscii(true)). The new test fails on the snake_case version and passes after the fix — confirmed by toggling locally.

Test plan

• WEAVIATE_VERSION=1.37.2 mvn verify -Dit.test=TextAnalyzerITest -Dmaven.javadoc.skip=true — passes against the patched record, fails 2/2 when the @SerializedName values are reverted to snake_case
• Full docs Java v6 suite (pytest -m java_v6 -k "Tokenization or SearchProfile") — green against the patched SNAPSHOT

🤖 Generated with Claude Code

[orca-security-eu]

Orca Security Scan Summary

Status	Check	Issues by priority
Passed	Infrastructure as Code	0   0   0   0	View in Orca
Passed	SAST	0   0   0   0	View in Orca
Passed	Secrets	0   0   0   0	View in Orca
Passed	Vulnerabilities	0   0   0   0	View in Orca