The autocomplete's `_updateSuggestions` method responds to user typing
input and queries for new suggestions based on it. However, since the
suggestions are updated after each keystroke (after the threshold), and
since the suggestion query typically fires an XHR, it's possible to
create an enormous number of suggestion query requests by typing
quickly. Depending on the hosting environment, this can lead to the user
exhausting their quota and getting stuck with HTTP 429 responses for a
while.
Add the an option to gr-autocomplete for a configurable debounce window.
Setting this to a relatively small value (200ms) in gr-search-bar and
the project lookup in gr-watched-projects-editor dramatically reduces
the number of requests from normal typing speed, putting HTTP-429-level
volume out of reach.
Bug: Issue 8149
Change-Id: Ie597189d1cd1575e686ab0c0041d4001ed8b6484
865bf1dd88