Updated smoke tests

[robgruen]

This pull request introduces several improvements to the test infrastructure and test data for the default agent provider, mainly focusing on normalizing URL comparisons in tests, refining test expectations, and improving test stability. The changes ensure that tests are less brittle in the face of minor output variations and that only the intended agents are evaluated during translation tests.

Test infrastructure improvements:

• Added normalization logic to treat bare hostnames and their http:///https:// forms as equivalent in test comparisons, reducing false negatives due to URL formatting differences (translateTestCommon.ts). [1] [2]
• Disabled "flow-only" agent schemas (e.g., utility) during translation tests to prevent them from interfering with the agents under test (translateTestCommon.ts). [1] [2] [3]

Test data and expectation refinements:

• Updated multiple test cases to use the anyof structure, allow alternate phrasings, and remove unnecessary parameters (like "tab": "current"), making tests more flexible and realistic (translate-browser-e2e.json, translate-e2e.json, translate-history-e2e.json, translate-gate-e2e.json, translate-image-history-e2e.json). [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17]

Workflow and CI stability:

• Improved the Azure pipeline's handling of Xvfb to prevent CI hangs by fully detaching its stdio and ensuring it is killed on exit (pipelines/azure-smoke-tests.yml).
• Commented out the explicit environment: development-fork assignment in the GitHub Actions workflow, possibly to restore conditional environment selection (.github/workflows/smoke-tests.yml).

UI bugfix:

• Fixed a CSS issue where the .conversation-name-bar would not be hidden when the hidden attribute was set, by explicitly setting display: none (chat.css).