Implement real link usage stats and debugMode logging

[dmccoystephenson]

Summary

• PersistentData.getTotalUses() and getMostPopularLink() were hardcoded to -1 and "(TBD)". These are now computed from real per-link use counts, and ViewCommand increments a link's use count on each successful lookup. The zero-links case is handled by returning "N/A" for the most popular link.
• The debugMode config option was documented in CONFIG.md as enabling console logging but had no effect anywhere in the codebase. A debug(String) helper was added to EasyLinks, gated on debugMode, and is now called from ConfigService.setConfigOption and from StorageService's save/load paths.
• CHANGELOG.md was updated with an Added entry describing both changes.

An unrelated pre-existing gap was noticed during this work: StorageService.save() is never invoked anywhere in the codebase (not even by CreateCommand/DeleteCommand), so link data — including use counts — is not currently persisted across restarts. That gap is outside the scope of the two issues closed here, so a separate tracking issue is being filed rather than folding a fix into this PR.

This PR description was drafted during a Gardener session (https://github.com/Stephenson-Software/gardener).

Test plan

• mvn compile — clean build
• mvn test — 8/8 tests pass, including 4 new tests covering getTotalUses and getMostPopularLink (empty-links and multi-link cases)
• Manual trace of ViewCommand.execute confirms link.setUses(link.getUses() + 1) runs only on a successful lookup, before the label/URL is sent to the player

Closes #9
Closes #8

[dmccoystephenson]

Self-review rubric:

• Scope: PASS — every modified file (EasyLinks.java, ViewCommand.java, PersistentData.java, ConfigService.java, StorageService.java, CHANGELOG.md, PersistentDataTest.java) is necessary for StatsCommand shows placeholder '-1' / '(TBD)' output for uses and most popular link #9 (usage stats) or debugMode config option documented as enabling logging but has no effect #8 (debugMode logging); no unrelated formatting or renames were included.
• Tests-new: PASS — the two changed public methods (getTotalUses, getMostPopularLink) each have 2 new tests (empty-links case and multi-link case); ViewCommand's use-count increment could not get a dedicated command-level test because this project has no mocking library on the test classpath (only plain JUnit 5) to fake a Bukkit CommandSender, so its correctness is covered indirectly through the PersistentData/Link semantics it relies on.
• Tests-fix: PASS — reverted PersistentData.java to its pre-fix state (git checkout HEAD~1 -- <path>) and re-ran PersistentDataTest: the 4 new tests failed with the old placeholder values (-1, (TBD)), confirming they are not false negatives. Restored the fix and re-ran: all 8 tests pass.
• Sibling structure: PASS — new tests follow the existing @BeforeEach/method-under-test-first naming convention already used in PersistentDataTest.
• Sibling renames: N/A — no parallel-pair identifiers were renamed in this PR.
• Docs: PASS — CONFIG.md's existing debugMode description ("Enables verbose debug logging to the console") is now accurate; COMMANDS.md/USER_GUIDE.md's generic "View statistics" wording for /el stats required no change; CHANGELOG.md gained an Added entry for both changes.
• Issue resolution: PASS — StatsCommand shows placeholder '-1' / '(TBD)' output for uses and most popular link #9's named surface area (PersistentData.getTotalUses(), getMostPopularLink(), and the never-called Link.setUses) is now implemented and exercised; debugMode config option documented as enabling logging but has no effect #8's named surface area (debugMode having no effect) is now wired to real logging via EasyLinks.debug(...).
• CI: PASS — the build check is green on the PR head SHA.
• Permissions: N/A — no permission node was added or changed; plugin.yml is untouched.

One pre-existing gap noticed but intentionally left out of scope: StorageService.save() is never invoked anywhere in the codebase, so no link data (including use counts) is currently persisted between restarts. This predates this PR and affects CreateCommand/DeleteCommand too, not just the stats feature — a separate tracking issue is being filed for it rather than expanding this PR's scope.