Deprecate the HTTP1 query endpoint

[agavra]

As of #4069 we have introduced a new /stream-query endpoint that leverages improved mechanics and HTTP2. Maintaining multiple such endpoints is error prone (as evidenced by #7109 and comments on #6944) so we should deprecate and remove the endpoint.

The problem is that (1) there are live use cases of that endpoint (e.g. the CLI, pull query-forwarding) and (2) lots of works has been put into optimizing /query for pull queries.

#6269 for some context on the Java Client
#6970 will also need to be tackled

[guozhangwang]

Just for my understanding: besides the CLI and the pull query forwarding, what are other use cases of the /query endpoint at the moment? I'd want to have some sense on how much investment we would need for item 1) before we can finally phase out this endpoint.

[agavra]

@guozhangwang I believe those are the internal use cases, there may be external use cases that we're not aware of. Though @AlanConfluent brought up that there's a lot of work bringing the new endpoint to performance parity to /query which may raise the LOE estimate.