sql: grow stack for DistSQL goroutines

[yuzefovich]

This commit makes it so that the main goroutine of Outbox and the vectorized hash router goroutine now start out with the larger stack.

Addresses: #130663
Epic: None
Release note: None

[blathers-crl]

It looks like your PR touches production code but doesn't add or edit any test code. Did you consider adding tests to your PR?

_{🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf.}

[cockroach-teamcity]

This change is 

[yuzefovich]

These goroutines previously showed up noticeably (like here), but not as much anymore in sysbench (probably because we improved distsql heuristics). I tried to coerce the sysbench microbenchmark to show any impact (using distsql=on and creating the tables on n2 while running queries from n1), but there was no change. Still, it makes sense that we'd need to grow the stack for these goroutines - we can do lots of execution processing inside, and since we're not using the stopper to start them, we don't benefit from 1aab60e. So this change seems sound to me, but wouldn't move the needle much.

(Initially I also adjusted growStackCodec to handle SetupFlowRequest in addition to kvpb.BatchRequest but decided to back out that change.)

[mgartner]

Thanks for following up on this!

[yuzefovich]

TFTR!

bors r+

[craig]

Build succeeded:

• lint
• linux_amd64_build
• linux_amd64_fips_build
• acceptance
• check_generated_code
• unit_tests
• local_roachtest_fips
• examples_orms
• docker_image_amd64
• local_roachtest