What problem does this PR solve?

Issue Number: related to #2199, #2322

Problem Summary:

After #2229, we can get the max snapshot lifetime by running SQL on TiDB[1]. But it is inconvenient for checking.

[1]

select tiflash_instance,max(STORAGE_STABLE_OLDEST_SNAPSHOT_LIFETIME),max(STORAGE_DELTA_OLDEST_SNAPSHOT_LIFETIME),max(STORAGE_META_OLDEST_SNAPSHOT_LIFETIME)
from information_schema.tiflash_tables
where 1=1 and (STORAGE_STABLE_OLDEST_SNAPSHOT_LIFETIME > 600 OR STORAGE_DELTA_OLDEST_SNAPSHOT_LIFETIME > 600 OR STORAGE_META_OLDEST_SNAPSHOT_LIFETIME > 600) 
group by tiflash_instance;

What is changed and how it works?

• Log query_id, read_tso in DMVersionFilterBlockInputStream deconstructor so that we can track a query by logging more clear
• Add a failpoint FailPoints::pause_after_copr_streams_acquired to mock that snapshot are not released by the copr level
• Each time AsynchronousMetrics::update run, it will collect the max oldest (stable/delta/meta) snapshot lifetime for each TiFlash instance and report it to Prometheus. Show them in the MVCC snapshots panel.
  

Related changes

• PR to update pingcap/docs/pingcap/docs-cn:
• Need to cherry-pick to the release branch:

Check List

Tests

• Manual test (add detailed scripts or steps below)

Side effects

Release note

• No release note