i - I case convertion with `PropertyNamingStrategy`

[cowtowncoder]

After fixing #953 -- issue with case-insensitive deserialization and "Turkish / dotted/undotted I" -- there is the remaining issue of proper handling of casing for deriving property names from accessors (getters / setters). It seems likely that matching would not work to connect fields with accessors (and/or explicit names) without passing Locale to allow for safe String.toLowerCase.

First part would be coming up with test case(s) to show the issue itself.

Earlier PR #2554 may be of use as well.