Fix per-index evaluation of ONE_FAILED in mapped task groups (#67684)

A task using a "fast triggered" trigger rule (ONE_FAILED, ONE_SUCCESS,
ONE_DONE) inside a mapped task group was evaluated against every expanded
instance of its upstream, instead of the upstream instance sharing its own
map index. As a result a single failed (or succeeded) upstream instance
wrongly triggered the rule for every expanded instance of the task — e.g. a
mapped ONE_FAILED reporting task ran for all map indexes when only one
upstream had failed.

The broad "depend on every upstream instance" behavior is only needed for
the not-yet-expanded summary task instance (map_index < 0), so a fast
trigger rule does not prematurely skip the task before the mapped task group
expands (the case fixed in #34023). Restrict that special case to the
summary instance; expanded instances now use the normal per-map-index
upstream resolution.

closes: #50210