feat: add support for Antigravity log parsing and analysis

[SecH0us3]

This PR introduces full local offline analysis support for Antigravity (Google DeepMind's agentic AI coding assistant) into the AI Engineer Coach extension.

Previously, Antigravity trajectories were not parsed or supported. With this PR, the extension now:

1. Discovers Antigravity Workspaces: Automatically discovers .db trajectory files from ~/.gemini/antigravity/conversations, ~/.gemini/antigravity-cli/conversations, and ~/.gemini/antigravity-ide/conversations
2. Parses Protobuf Payloads: Reads the SQLite-backed trajectory databases and decodes the protobuf step payloads to map them to the standard Session and SessionRequest format.
3. Dynamic Model Detection: Accurately extracts the underlying model ID directly from the protobuf payload (supporting gemini-, claude-, gpt-, and o1/o3- variants), ensuring correct model usage is tracked in the Output dashboard
4. Skills & Code Extraction: Extracts the paths of skills used from the trajectory messages and maps them to standard skill names. It also parses code blocks and tracks image usage to properly populate the Output dashboard and Skill Finder workflows

This allows users to get deep insights into their Antigravity agentic sessions directly within the VS Code dashboard without telemetry or sending logs externally

[SecH0us3]

@microsoft-github-policy-service agree

[SecH0us3]

@aymenfurter Good day! Could you please take a look at my PR when you have a moment?
I really would like to have the feature and I'm open to any changes