[SPARK-18360][SQL] default table path of tables in default database should depend on the location of default database

## What changes were proposed in this pull request?

The current semantic of the warehouse config:

1. it's a static config, which means you can't change it once your spark application is launched.
2. Once a database is created, its location won't change even the warehouse path config is changed.
3. default database is a special case, although its location is fixed, but the locations of tables created in it are not. If a Spark app starts with warehouse path B(while the location of default database is A), then users create a table `tbl` in default database, its location will be `B/tbl` instead of `A/tbl`. If uses change the warehouse path config to C, and create another table `tbl2`, its location will still be `B/tbl2` instead of `C/tbl2`.

rule 3 doesn't make sense and I think we made it by mistake, not intentionally. Data source tables don't follow rule 3 and treat default database like normal ones.

This PR fixes hive serde tables to make it consistent with data source tables.

## How was this patch tested?

HiveSparkSubmitSuite

Author: Wenchen Fan <wenchen@databricks.com>

Closes #15812 from cloud-fan/default-db.

(cherry picked from commit ce13c267)
Signed-off-by: Yin Huai <yhuai@databricks.com>