Clean up HoodieSparkSqlWriter

HoodieSparkSqlWriter has evolved over the years w/ layer on layers of logic being stacked in it. It's reaching a point where it's virtually impossible to understand interactions and number of permutations b/w different features. One recent example is HUDI-4472.

We need to make sure that it's in a state where:
# Writing logic is easily traceable, and it's possible to understand how different features might impact the write-path.
# Make sure that it's structured in a way allowing plugging in new features/revisiting existing logic (for ex, allowing different path for row-writing not only for bulk-insert but other operations as well, etc)
#  

## JIRA info

- Link: https://issues.apache.org/jira/browse/HUDI-4692
- Type: Improvement
- Epic: https://issues.apache.org/jira/browse/HUDI-1297

Start by locating and reading HoodieSparkSqlWriter, then use HUDI-4472 and the linked HUDI-1297 epic to map the current write-path permutations. Done means the writer's logic is traceable and structured for the feature-path needs described in the issue; no specific test is named in the payload.

Written by the indexing model from the issue text.