We are deferring the CLDR intervalFormats change for this to CLDR 1.8, so this bug goes to ICU 4.4
The goal is:
1. if the Key has 'hM' only and Data is 'hM',
input hM, or HM should match either input and produce the hM data
2. if the Key has both 'hM' and 'HM',
and Data is 'hM' and 'HM',
Input hM should match hM data, BUT input 'HM' should match HM data
Should use same pattern matching as DateTimePatternGenerator ("closest match")
reassigning to pedberg provisionally.
Checked in some changes to address thus. Have not fully expanded the tests to check this yet, since the new tests will depend on the new CLDR data that is not in yet. The current tests only check skeletons with 'h' and they pass.
Milestone 4.3.5 deleted