test: add Unicode and large blob storage regression test

[saltus7]

Description

Summary

Adds two self-contained regression test suites targeting edge cases in the single-file storage layer. No production code changes.

What's included

UnicodeStorageTests.cs (8 tests)

• 3-byte CJK roundtrip (Japanese, Chinese, Korean) with WHERE clause matching
• 4-byte emoji including ZWJ sequences (family emoji, skin tone modifiers)
• RTL scripts (Arabic, Hebrew)
• Combining character equivalence — verifies the engine performs byte-exact comparison without normalization (precomposed é vs decomposed e+́ are treated as distinct)
• Large mixed-script payload roundtrip at the block level

LargeBlobCompressionTests.cs (8 tests)

• Roundtrip integrity for 1KB, 64KB, 1MB, and 16MB payloads using SHA256 hash verification
• 16MB test forces ~4,000 page chains, validating block chaining and reassembly
• High-entropy (incompressible) data to verify compression fallback path
• Multiple large blobs in the same file to verify registry isolation

Motivation

These tests cover scenarios that are easy to get silently wrong:

• Unicode normalization bugs are invisible until a user queries café and gets no results because the database stored cafe+́
• Block chaining bugs on large blobs produce silent data corruption — the read succeeds but the bytes don't match
• Compression fallback failures only manifest on incompressible data, which is rarely tested

All tests use temp file cleanup via IDisposable and don't require any external infrastructure.

Test results

UnicodeStorageTests: 8/8 passed (1.15s)
LargeBlobCompressionTests: 8/8 passed (2.13s)

Breaking changes

None. Test-only PR.

Note: This PR includes the compression feature from PR #344 as a dependency. The test files reference BlockCompressionMode introduced in that commit. If the maintainer prefers, these tests can be merged together with the compression feature, or rebased onto master after #344 merges.

[sonarqubecloud]

Quality Gate passed

Issues
1 New issue
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud