refactor: give water slide modes their own enum values

[rohitpatil9121]

Summary

Fixes #987

WaterModes water slide members reused the value strings of the legacy modes ("low", "medium", "high", "extreme"). StrEnum deduplicates by value, so these members became aliases of LOW / MEDIUM / HIGH / EXTREME and lost their own codes. For example, WaterModes.PURE_WATER_FLOW_MIDDLE.code was 202 instead of 235, and WaterModes.from_code_optional(235) returned None.

This follows the approach suggested in the issue:

• Each water slide member now has a unique value, and the shared user-facing name moves to display_name, the same way CleanRoutes.DEEP_PLUS_CN works.
• The workarounds that existed only to keep the slide codes are removed:
  • the WATER_SLIDE_MODE_MAPPING side table, replaced by a private tuple of slide modes
  • the reverse lookup by name in get_cleaning_mode_parameters, which now uses water_box_mode.code directly
  • the inline isinstance patch in get_current_cleaning_mode, now _resolve_water_mode(value, features), matching _resolve_clean_mode
• get_water_mode_mapping and get_water_slide_mode_name use display_name, so the user-facing names and the mapping stay the same.
• A conformance test now fails when members of a RoborockModeEnum alias each other by value. B01Fault already has such aliases, so I added it to a baseline (strict xfail) instead of changing it here.

Behavior change

.value for the four slide modes is now pure_water_flow_small, pure_water_flow_middle, pure_water_super_begin and pure_water_flow_end. Consumers that read mode.value from water_mode_options should switch to mode.display_name. For Home Assistant, that is step 1 of the upgrade plan in the issue. water_mode_mapping and water_mode_name return the same strings as before.

Testing

• Added a test that the slide codes 225/235/245/250 resolve to their own members with the expected display_name.
• Updated test_water_slide_mode_mapping to read display_name from water_mode_options.
• pytest tests/devices tests/conformance tests/data: 1225 passed, 54 xfailed
• ruff check and ruff format --check pass. mypy reports no errors in the changed files.
• tests/test_web_api.py::test_url_cycling fails locally on Windows, on main as well, so it is unrelated to this change.