feat: Independent system/information_schema database in different catalogs

[sundy-li]

Each catalog should have its own system/information_schema database rather than keep them in default catalog

[TCeason]

This will greatly simplify the optimization work for querying certain system tables under the system. For example, tables are now placed in default.system.tables, and the optimization logic would be quite complex.

However, after the split, external users need to know which catalog they need to look at. This requires some adjustments from external callers.

For example, if I have catalogs ctl1, ctl2, and default, and I previously wanted to query a table named a, I would need to execute select * from default.system.tables where name=a.

After the catalogs are separated, the query would need to be select * from default.system.tables where name=a union all select * from ctl1.system.tables where name=a union all select * from ctl2.system.tables where name=a.

Perhaps some HTTP API interfaces may also need to be modified. Please be aware of this in advance. If there are any issues, they can be discussed in the issue section or after a PR.

cc @flaneur2020