kvserver: co-construct AllocatorSync in NewStore

[tbg]

Production code (server.go) pairs MMAllocator with an AllocatorSync on
the StoreConfig. A number of test entry points populate MMAllocator (via
TestStoreConfig) and StorePool but leave AllocatorSync unset. The MMA
store rebalancer worked around this with a buildutil-gated fallback that
lazily constructed an AllocatorSync in test builds and panicked otherwise.

That gate was tripping benchmarks, which are compiled without the crdb_test
tag, producing failures like:

panic: AllocatorSync must be set on StoreConfig outside of test builds

This PR co-constructs AllocatorSync inside NewStore when MMAllocator is
set but AllocatorSync is not, mirroring what server.go already does
explicitly. The fallback in newMMAStoreRebalancer is replaced with a plain
non-nil assertion, and the now-redundant explicit construction in
localtestcluster is dropped.

Verified locally with ./dev bench (which builds with --crdb_test_off,
matching the failing TC RoachprodMicrobenchmarkWeeklyWip configuration):

./dev bench pkg/kv/kvserver -f BenchmarkStoreGetReplica --bench-time=1x

Fixes #170455.
Fixes #170456.

Release note: None
Epic: none

[trunk-io]

😎 Merged successfully - details.

[cockroach-teamcity]

This change is 

[blathers-crl]

Detected infrastructure failure (matched: self-hosted runner lost communication with the server). Automatically rerunning failed jobs. (run link)

[angeladietz]

LGTM. out of curiosity, would it also have been an option to just specify the AllocatorSync in all the tests, to make the tests more representative of production code?

[blathers-crl]

Based on the specified backports for this PR, I applied new labels to the following linked issue(s). Please adjust the labels as needed to match the branches actually affected by the issue(s), including adding any known older branches.

---

Issue #170455: branch-release-26.2.

---

Issue #170456: branch-release-26.2.

---

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

[blathers-crl]

Encountered an error creating backports. Some common things that can go wrong:

1. The backport branch might have already existed.
2. There was a merge conflict.
3. The backport branch contained merge commits.

You might need to create your backport manually using the backport tool.

---

merge conflict cherry-picking c09e0f8 to blathers/backport-release-26.2-170638

Backport to branch 26.2.x failed. See errors above.

---

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

[tbg]

LGTM. out of curiosity, would it also have been an option to just specify the AllocatorSync in all the tests, to make the tests more representative of production code?

Yes, I looked at both and was on the fence (because I consider "test-only init paths in prod" a risk generally). But it was tedious to construct an AllocatorSync in ~12 places. Let me take another look though.