test: share Stovepipe MySQL integration stack

[sbalabanov]

Why?

The Stovepipe MySQL storage integration target started and destroyed three separate Compose stacks for the RequestStore, QueueStore, and BuildStore suites. Container startup and MySQL initialization dominated the target runtime.

What?

• Start one MySQL Compose stack and apply the storage schema once for the target.
• Run the existing RequestStore, QueueStore, and BuildStore contract suites serially against the shared storage instance.
• Truncate every storage table before each suite so fixed test identifiers and persisted state remain isolated.

Uncached Bazel test-attempt runtime across three runs improved from 45.273s / 40.420s / 40.358s (42.017s average) to 14.111s / 14.044s / 14.056s (14.070s average), a 66.5% reduction.

Test Plan

• ./tool/bazel test //test/integration/stovepipe/extension/storage/mysql:go_default_test --nocache_test_results (three post-change runs)
• make fmt
• make lint
• make check-tidy
• make check-gazelle

Issue

N/A

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}