feat: compute embeddings through qdrant

[Teagan42]

What

• remove direct fastembed model usage from server
• delegate search embedding to Qdrant Fastembed using Document inputs
• bump version to 0.26.1

Why

• use Qdrant's built-in embedding support instead of managing models manually

Affects

• mcp_plex/server.py
• tests/test_server_cli.py
• pyproject.toml, uv.lock

Testing

• uv run ruff check .
• uv run pytest

Documentation

• n/a

---

https://chatgpt.com/codex/tasks/task_e_68c57e90bdc0832890da728a3a59f533

[github-actions]

Coverage Report

File	Stmts	Miss	Cover	Missing
mcp_plex
loader.py	196	15	92%	32–34, 306, 309–312, 357–366, 652
server.py	220	12	95%	79, 97–98, 217, 445, 449, 469–473, 520
TOTAL	519	27	95%

Tests	Skipped	Failures	Errors	Time
25	0 💤	0 ❌	0 🔥	29.779s ⏱️