fix: lower max batch size from 2000 to 200

[cskmnrpt]

Lowers the batch-size ceiling in qase-python-commons from 2000 to 200.

What changed

• qase-python-commons/src/qase/commons/models/config/batch.py — set_size rejects a size above 200. The old message said "less than 2000" while the check tested > 2000; the new wording matches the check.
• qase-python-commons/src/qase/commons/reporters/testops.py and testops_multi.py — clamp with min(200, ...).
• qase-pytest/docs/UPGRADE.md and qase-robotframework/docs/UPGRADE.md — state the 1-200 range.

The default stays 200, so no user on defaults is affected.

Tests

qase-python-commons suite: 243 passed, 4 skipped.

Why

TMS-196 lowers the bulk-results cap from 2000 to 200 results per request. Very large batches arrive as a single spike in the result-processing queue and delay processing for every customer, including runs that report a handful of results at a time. Fewer than 1% of batches exceed 200 today.

This PR removes the stale 2000 from this repository. Server-side enforcement lands separately in the app and specs repositories.

Flags for the reviewer

1. Missing clamps. Only Java and Python validate the configured batch size at all. JavaScript, C#, Go, qasectl and PHP pass a configured value straight to the API, so a user who set 500 gets an HTTP 413 and lost results instead of a local warning. Adding a clamp is a separate call — your decision whether and when.
2. Generated API clients still say 2000. The maxItems: 2000 constraint and the "more than 2,000 results" endpoint description are generated from the v1 OpenAPI spec into every client (qase-api-client, qase-php-client, and the vendored clients in this and the other language repos). They change by regenerating from the spec, not by hand. Note qase-javascript carries two generated trees (qase-api-client/ and qaseio/src/generated/), so a partial regeneration leaves one stale.

Full cross-repo inventory of every mention, including an org-wide sweep of all 163 qase-tms repositories, is available on request.