fix flaky recommender rerank_depth test

[danyilq]

• Please check if the PR fulfills these requirements

• The commit message follows our guidelines
• Tests for the changes have been added (for bug fixes/features)
• Docs have been added / updated (for bug fixes / features)

• What kind of change does this PR introduce? (Bug fix, feature, docs update, ...)

  • Bug fix
  • Documentation update
  • Minor test behavior improvement

• What is the current behavior? (You can also link to an open issue here)

  • Some integration tests in py-marqo, particularly for recommender and facets, exhibit flakiness and inconsistent cleanup behavior.
  • The create-indexes=True parameter in cloud tests causes existing indexes to be deleted, which makes tests more annoying to deal with.
  • Test setup instructions are scattered and unclear.

• What is the new behavior (if this is a feature change)?

  • Test flakiness related to recommender and facets tests is resolved by analyzing the roor cause and excepting different values depending on environment since cloud tests are multi-node.
  • One test in test_tensor_search has been updated for improved readability.
  • A new README.dev.md has been added under the tests/ directory, documenting how to properly run integration tests against both Open Source and Cloud environments, including GitHub Actions usage and cleanup procedures.
  • The create-indexes flag has been updated: if set to True and the indexes already exist, they are now skipped rather than deleted, allowing tests to proceed without destructive actions.

• Does this PR introduce a breaking change? (What changes might users need to make in their application due to this PR?)

• Other information: