Returns the limits actually enforced for (computeNodeId, userId),
with per-axis source attribution and current per-period spend.
Mirrors the resolution chat-service runs at dispatch time:

per-user row → node __default__ row → platform safety cap

axis-by-axis (any unset axis falls through to the next tier). Use
this rather than the raw /user-quotas/{userId} GET when you want
to know what's actually applied — the raw GET returns 404 when no
per-user override exists, which is misleading because the user is
still bound by __default__ and the safety cap.

Pass me as userId to retrieve your own effective quota.

Permissions: Node owner / workspace admin can read any user;
the caller can also read their own effective quota.