chore: migrate to ESLint config v3

[kpal81xd]

ESLint config v3 migration

Part of playcanvas/eslint-config#48.

Scope

• Rebased the existing PR branch onto current main and rebuilt the stale migration.
• Migrated to @playcanvas/eslint-config 3.0.0-beta.8 via /typescript, removing direct parser, plugin, and resolver dependencies now supplied by the shared config.
• Added the shared Prettier config, fmt/fmt:fix/typecheck scripts, and CI coverage for formatting, linting, type-checking, tests, build, and publint.
• Applied shared formatting and replaced explicit any with sound metadata, worker, device, logger, and serialization types without shipped runtime behavior changes. viewerSettingsJson intentionally tightens from any to unknown.
• Made render-golden tests skip only when no WebGPU adapter is available, matching the existing GPU-test behavior on headless CI.

Behavior proof

• Whole-branch executable comparison found no shipped runtime behavior change; ownership checks preserve main method lookup/receiver semantics, and no bare side-effect imports were introduced.
• The ESM runtime export list is unchanged; logger inputs remain call-compatible and viewerSettingsJson is the only root public declaration tightening.
• The two local GPU render-golden failures are baseline-only: an isolated origin/main build produced the exact same mid and tiny-dof results on this machine.

Verification

• git refresh — stale one-commit migration conflicted with current main; the obsolete commit was skipped and the migration rebuilt on main, then force-pushed with lease.
• npm ci — passed; 423 packages audited, with 5 reported dependency vulnerabilities.
• npm run fmt — passed.
• npm run lint — passed.
• no-explicit-any audit — zero violations, disables, or suppressions; seven avoidable no-empty-function directives were removed.
• npm run typecheck — passed.
• npm run build — passed.
• npm test — 794 of 796 passed; an isolated origin/main build reproduces the exact two GPU render-golden failures. Focused affected suites pass 86/86.
• node --import tsx --test test/render-golden.test.mjs against isolated origin/main — reproduced the same 2 local golden failures.
• npm run publint — passed with one sideEffects suggestion.
• npm run docs — passed with 0 errors and 11 documentation-link warnings.
• npm pack --dry-run — passed; 217 files packed.
• git diff --check — passed.
• GitHub CI — Build, Lint, and Docs passed; the headless runner skips only the 3 WebGPU render-golden cases.

Manual smoke test

1. Run splat-transform on an existing PLY input and write a PLY output.
2. Confirm the CLI completes and the output opens with the same splat count and appearance.