docs(skill): document context_length auto-detection resolution chain

[Railway9784]

When model.context_length is set in config.yaml, it blocks auto-detection from the server's /v1/models endpoint. The skill incorrectly implied a hard fallback to 131072. Add the resolution chain and the fix command (hermes config set model.context_length "") to both the config table and a new troubleshooting section.

What does this PR do?

The bundled hermes-agent skill's config table and troubleshooting
section don't explain that model.context_length blocks auto-detection
from the server's /v1/models endpoint, or how to restore it.

Hermes uses a 4-step resolution chain:

1. model.context_length in config.yaml (blocking if set)
2. Custom provider per-model
3. Persistent cache
4. /v1/models auto-detect

Two changes:

• Config table: note that clearing the key enables auto-detect
• New troubleshooting section: resolution chain + hermes config set model.context_length "" fix

Real-world case: user had -c 262144 but Hermes showed 128k because a stale config value survived from a previous setup.

[teknium1]

Merged via PR #49914 (docs salvage cluster). Your commit was cherry-picked onto current main with your authorship preserved in git log — merged with rebase so it lands under your name. Thanks for the contribution!