rust(docs): warn that unscoped list_channels name searches can time out

[alexa-sift]

Description

In large orgs, list_channels name searches without an asset_id scope scan every channel and can hit the client request timeout. Agents then retry the same broad filter. The list_channels guidance now says to resolve candidate assets with list_assets first, search each asset by name, and treat a timeout as a too-broad filter rather than bad parameters.

Verification

• cargo fmt -p sift_mcp -- --check, cargo clippy -p sift_mcp --all-targets --all-features (no new warnings)
• cargo test -p sift_mcp (581 pass), including a new test that the list_channels description carries the timeout guidance
• Scripted tools/list check: ran sift-cli mcp from a local build of this branch (placeholder profile, no backend calls) and read the list_channels description over stdio. All 42 tools listed. The new bullet comes through with " unescaped and no \" or literal \n. Diffed against sift-cli 0.7.2, the only change is the three new lines:

  - Unscoped name searches scan every channel in the org and can time out in large orgs. Resolve
    candidate assets with `list_assets` first, then search each with `asset_id == "..." && name.matches(...)`.
    A timeout means the filter was too broad, not that the parameters were wrong; narrow it before retrying.

[bijan-sift]

is there a way to confirm this works?

[alexa-sift]

is there a way to confirm this works?

@bijan-sift fair question. The linked azimuth PR has more integration-type tests that I originally included, but I added a test here as well. Lmk your thoughts