[FEA] Revert changes to nvcomp decompression status check in parquet

[devavret]

First discussed here: #11064 (review)

#10699 made a change to the decompression status check that reverted the changes made in #9995.
The latter added a custom kernel to check statuses to avoid using thrust::all_of which supposedly caused a device synchronize and deteriorated perf when running multiple parquet read calls on the same device.

This issue is to request the reverting of this particular change https://github.com/rapidsai/cudf/pull/10699/files#diff-69acbfcb0f4330a9fc3feec2aca0ceab2da0715fc1d7831304c2b154fb1a91e5L1053-L1083

[jbrennan333]

I talked to @cheinger about this at our sync up today. He mentioned TPC-DS query 9 as one of the ones that was most affected by his optimization. Looking back at our spark-rapids weekly benchmarks, we don't see any degradation in q9 since #10699 went in. Based on what he said, it is not clear that we need to revert this change.

[github-actions]

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.

[jbrennan333]

@devavret, @vuule, do you agree this can be closed?