Some remaining tasks for spider select handler V1 implementation

(copied from https://jira.mariadb.org/browse/MDEV-37202?focusedCommentId=309320&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-309320)

Further tasks (denote V1 as the version that will be pushed as MDEV-27260, and V2 as a subsequent version) as described in the design doc (see the heading "Next steps"):

• [X] For derived handlers and subqueries, leave that to V2.
• [ ] For test coverage of sql query results, transform main suite tests such as main.select to use spider. Requires some light text editing scripting. Do this in V1 if possible.
• [X] For further test coverage including testing using different configurations for crash or wrong results, I believe this is something that can be handled by testers
• [ ] For gbh coverage dropping caused by this change, use rdiff and spider_disable_select_handler combination. Do this in V1.
• [X] SELECT modifier optimizer hints (HIGH PRIORITY, SQL_CACHE, SLQ_BUFFER_RESULT, SQL_SMALL_RESULT, SQL_BIG_RESULT, STRAIGHT_JOIN, SQL_CALC_FOUND_ROWS, USE/FORCE/IGNORE INDEX): ignore if possible in V1. Support in V2.
• [X] xa: disable sh if in xa in V1. Support in V2.