fix(id): address Santanu's review feedback on #109

[r0goyal]

Summary

Addresses actionable items from Santanu's review on #109.

Breaking-change concerns

• IdValidationConstraint.isValid(Id) -> isValid(InternalId) (binary-breaking): restored isValid(Id) as the interface's abstract (compat-preserving) method. isValid(InternalId) is now a default method that bridges via new IdValidationConstraint.toId()/toInternalId() helpers, so existing external implementers keep compiling. PartitionValidator now overrides isValid(InternalId) directly (its natural contract) and implements isValid(Id) via the bridge.
• IdGenerator.initialize(int, List, Map) removal: confirmed intentional, no change.
• GenerationResult: documented as internal-only (Javadoc) - it's used across the generator/nonce subpackages so true package-private visibility isn't feasible without a larger restructure, but it's not meant to be treated as public API surface.

Design/correctness

• IdGeneratorType.LOOKUP_MAP keyed by IdFormatter object identity: added IdFormatter.id() (stable string identifier implemented by DefaultIdFormatter/RandomNonceIdFormatter) and rekeyed the lookup map off that instead of raw instance identity. Closes the fragility completely - works even with a freshly constructed formatter of the same kind, not just the cached singletons.
• IdParsersV2 DEFAULT_PATTERN ambiguity: added a comment explaining the two-step V1/V2 disambiguation (match, then require group(3) empty).
• NodeUtils vs IdGeneratorBase re-set inconsistency: left as-is per discussion.
• Dual registration burden (constraints registered separately on IdGenerator vs IdGeneratorV2): left as-is - callers using IdGeneratorV2 are expected to initialize it explicitly/independently.

Minor/style

• Added missing Apache 2.0 license headers to the 9 flagged new files (+ GenerationResult.java).
• Removed stale @SneakyThrows from IdGeneratorPerfTest/IdGeneratorPerfV2Test benchmark methods (no checked exceptions thrown).
• Removed unused IdGeneratorBase import in IdParsersV2Test.
• IdGenerationRequest (dead - no longer used by either generator) marked @deprecated with an explanatory Javadoc rather than removed outright, to avoid a silent source break for any external caller.

Tests

• Added IdGeneratorV2Test#testV2IdParseWithLegacyGeneratorDoesNotRoundTrip, documenting the (lossy/asymmetric) interop contract: a legacy id parses fine via IdGeneratorV2.parse(), but a V2-generated id does not round-trip through the legacy IdGenerator.parse(), since it has no knowledge of the embedded IdGeneratorType code.
• IdGeneratorV2.cleanUp() and IdGenerator.cleanUp() now both call NodeUtils.reset() (applied symmetrically since NodeUtils is shared state across both generators) - closes the latent footgun where a stale node id could be silently reused by a subsequent initialize().

Verification

• mvn -pl ranger-discovery-bundle -am clean test - all test classes pass (0 failures/errors).

Base for review: this PR targets id-v2 (not main) so the diff can be reviewed/approved before folding into the main id-v2 branch feeding #109.

[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