See test data StartsWith_StringComparison_TestData. "\u00ADHello" and "Hello" is expected to be treated the same and as the consequence, "Hello".StartsWith("\u00ADHel") is expected to return true.
Should we add soft hypen to normalization step and ignore it for StringComparison.CurrentCulture? If we cannot do anything about it, add it to known platform differences in the docs.

The failing tests are using mono_wasm_index_of.