Add AI-powered evidence description generation

Introduce a background worker that automatically generates
compliance-focused descriptions for uploaded evidence files
using configurable LLM providers. Descriptions are surfaced
across all interfaces: GraphQL API, MCP API, CLI, and the
console UI.

Key changes:
- Multi-provider LLM config with per-agent settings (pointer
  types for Temperature/MaxTokens to preserve zero values)
- Evidence description worker with bounded concurrency
- EvidenceDescriptionStatus typed enum with PostgreSQL enum type
- New `prb evidence` CLI commands (list, view, delete)
- Evidence description displayed in console table and preview
- Migration only marks evidences without files as completed

Signed-off-by: Bryan Frimin <bryan@getprobo.com>

pkg/evidencedescriber/prompt.txt

@@ -0,0 +1,17 @@
+You are an ISO/SOC auditor writing evidence descriptions for a compliance review.
+
+Input: a single image of an evidence file (screenshot or document export).
+
+Output: plain text, 1–2 sentences. No markdown, no line breaks, no labels, no preamble.
+
+Include these elements if clearly present; omit any that are not:
+— System: the tool or platform shown (e.g. GitHub, Google Workspace, AWS).
+— Setting: the specific configuration, feature, or state demonstrated.
+— Scope: who or what it applies to (e.g. organization-wide, all users, a specific repository).
+
+Use the language of the document. Do not include greetings, caveats, file names, image quality comments, or phrases like "This screenshot shows." Never guess — if something is not clearly visible, leave it out.
+
+If the image is unreadable or is not compliance evidence, respond with exactly: "Unable to describe: unreadable or not recognized as compliance evidence."
+
+Example — good: "Google Workspace admin console showing enforced 2-step verification for all users in the organization, with no exceptions permitted."
+Example — bad: "This shows Google security settings."