ENH: serialize scipy sparse arrays (sparray / dok_array)

[charan-rathore]

Closes #9353

The spmatrix classes are deprecated in favour of sparse arrays, and the array types don't inherit from spmatrix, so they currently get no serialization handlers - transmitting a dok_array (or any other sparray) fails with TypeError: Could not serialize object of type dok_array. This is what reds the nightly CI (14 failures in distributed/protocol/tests/test_scipy.py).

This registers generic dask serialization for scipy.sparse.sparray and adds dok_array handlers mirroring the dok_matrix ones, and parametrizes the scipy protocol tests over both the legacy matrix classes and the modern array classes.

Verified locally: without the scipy.py change the new array parametrizations fail 14/28; with it, 28/28 pass. Ruff clean.

• Tests added / passed
• Passes pixi run lint (ran ruff directly on the two changed files - clean; no pixi on this machine)

[github-actions]

Unit Test Results

See test report for an extended history of previous test failures. This is useful for diagnosing flaky tests.

    40 files  ±  0      40 suites  ±0   14h 8m 30s ⏱️ - 6m 26s
 4 174 tests + 14   3 993 ✅ + 15    178 💤 ±0  3 ❌  - 1 
81 198 runs  +241  76 953 ✅ +238  4 241 💤 +3  4 ❌ ±0 

For more details on these failures, see this check.

Results for commit 03d882a. ± Comparison against base commit dc182bd.

[charan-rathore]

CI verdict on the 3 failures: all three are known flakes, unrelated to this change.

• test_contact_listen_address[tcp://127.0.0.2---nanny] (1/20 runs) — tracked in Flaky test_contact_listen_address #6742.
• test_closed_input_only_worker_during_transfer (2/18 runs) and test_workers_do_not_spam_get_requests (1/18 runs) — shuffle flakiness tracked in Flakiness in test_shuffle.py #8074.

This PR only touches scipy sparse serialization and its protocol tests; those files are untouched by the failures above, and the full scipy protocol suite (matrix + array classes) passes. Nothing to change on my side.

[charan-rathore]

Closing this one. It hasn't had maintainer traction since I opened it and I'm focusing my effort elsewhere. Happy to revisit if there's interest.