* Follow HTTP 307 redirects when POSTing a query to Presto

This is how the Presto team communicated on IRC to properly "dispatch"
queries to Presto clusters, to enable replacing a cluster without
impacting users of the cluster.

* Ensure we talk to the correct Presto coordinator when cancelling a query

The previous behaviour would build a cancellation URI from the query ID
and the configured host (unless SSH tunneling is configured). This
commit builds on that existing behaviour, replacing the host and port
in the URI with ones extracted from the `infoUri` in the polling
responses, but only if they match the ones configured in the datasource.
This should not impact connections going over an SSH tunnel as sending
the request over the SSH tunnel requires using a different host than the
one configured.

Test data loading improvements 

New joins syntax

[ci oracle]

[ci oracle]

[ci drivers]

Add identifier-type to hx/identifier 

[ci drivers]

Code cleanup 

Test suite improvements & other code cleanup

Async QP improvements 

Faster test launch

Add :joined-field clause to MBQL

fix can not expand string when page too short and text too long

Fix for Custom Field as filter

Fixed #9250: Auto Resize Native Query Results

Add Identifier type to support all special chars in SQL names 

[ci drivers]

Merge 0.32.8 -> master

Reschedule Quartz tasks on launch

Fix occasional test failures because a timeout was too short 

Use delay for cache-backend-instance so there is no thrashing on init

Fix admin Settings updates page before version info is fetched 

Fix sql-jdbc QP error responses not being properly reported