Add Playwright task record/playback commands with TaskOrchestrator utilities

[NerioVillalobos]

Motivation

• Provide first-class support for recording and replaying manual procedures against a Salesforce org using Playwright integrated into the CLI.
• Centralize Playwright setup, dependency management and common error diagnostics into a reusable orchestrator so commands can surface actionable suggestions.
• Improve developer experience when working with a linked ESM plugin by documenting the required build step.

Description

• Add new commands metadelta:task:record and metadelta:task:play with full implementations in src/commands/metadelta/task/ and compiled outputs in lib/commands/metadelta/task/ to record Playwright codegen sessions and execute patched Playwright tests against an org.
• Introduce src/utils/task/orchestrator.js (and compiled lib/utils/task/orchestrator.js) implementing TaskOrchestrator, SF/NPX command helpers, Playwright dependency/browser installers, test file helpers (injectBaseUrlInTest, resolveTestFilePath, etc.) and solution hints for common failures.
• Add test helper routes and defaults under tests/ (metadelta-task-orchestrator-routes.js, metadelta-task-orchestrator.json) and wire new commands into the oclif.manifest.json.
• Documentation and build adjustments: update README.md with a linked-ESM note and npm run build instruction, add a prepare script and topicSeparator to package.json, and apply formatting/cleanup to postvalidate.js.

Testing

• Executed a TypeScript build via npm run build (which runs tsc and the postbuild manifest generator) and the build/manifest generation completed successfully.
• No automated unit tests were added for Playwright flows in this change set; Playwright sessions require runtime validation against a target org and were not executed as part of the automated build.

---

Codex Task