Improve API and configs for error Table

This is a follow up Jira to review comments in [https://github.com/apache/hudi/pull/7982/files]
The config names for error table can be generalised to accomadate for datasource and spark sql as well. Currently they are tied to DeltaStreamer. ([https://github.com/apache/hudi/pull/7982/files#r1118341097|https://github.com/apache/hudi/pull/7982/files#r1118341097)])
The API should throw exception so that any error cases can be handled. ([https://github.com/apache/hudi/pull/7982/files#r1118372238|https://github.com/apache/hudi/pull/7982/files#r1118372238)])
The constructor for error table writer can be generalised to handle DS and spark sql use cases as well. ([https://github.com/apache/hudi/pull/7982/files#r1118345162|https://github.com/apache/hudi/pull/7982/files#r1118345162)])

Add API for starting a commit. Handle concurrent error events or concurrent writes to the error table. ([https://github.com/apache/hudi/pull/7982/files#r1118350161|https://github.com/apache/hudi/pull/7982/files#r1118350161)])

Handle error records from writeStatusRDD ([https://github.com/apache/hudi/pull/7982/files#r1118380323])

## JIRA info

- Link: https://issues.apache.org/jira/browse/HUDI-5858
- Type: Bug

Start by reviewing the linked PR 7982 and its comments, then trace the error table configuration, API, writer constructor, commit-start flow, and writeStatusRDD handling. The work is done when the API and configuration support datasource and Spark SQL use cases, concurrent error-table writes are handled, and error cases are surfaced consistently.

Written by the indexing model from the issue text.