Content filter should identify the flagged word or phrase

Description

When submitting an issue via the CLI, the API content filter returns a generic error:

{"error":"Description contains inappropriate language"}

This gives no indication of which word or phrase was flagged, so the submitter has no way to fix the problem without guessing. In practice this leads to repeated trial-and-error resubmissions that pollute the issue tracker with junk.

Steps to reproduce

1. Submit an issue via swamp issue bug/feature with a body that triggers the filter
2. Receive the generic error with no detail about what was flagged
3. Have no way to know what to change

Expected behavior

The error response should identify the flagged term or at least the region of text that triggered the filter, e.g.:

{"error":"Description contains inappropriate language","flagged":"some phrase","position":{"line":12,"column":5}}

This lets the submitter fix the actual problem in one shot instead of blindly rewriting or, worse, submitting test issues to bisect the filter.