Update file change debounce to wait longer

When the host sees a file change, it waits 500ms for any further file changes to determine if it should defer restarting (because file changes might still be in progress). It seems like for some customer deployments, this wait period is too short, and multiple host restarts end up occurring for a single deployment. This leads to application errors (because of starting the host in a partially deployed state).

I propose we increase this to 1000ms. The downside is small (slightly more latency when editing a file and having the host pick it up) but the upside is more reliable deployments.

Deployment workflows such as run from zip and app_offline will be unaffected by this change.

Search the repository for the 500ms file-change debounce and inspect the host restart path that uses it. Update the wait to 1000ms and run the relevant existing tests, confirming that file changes no longer trigger multiple restarts during a deployment.

Written by the indexing model from the issue text.