[lld] [LLD] [MinGW] Sync --thinlto-cache-dir option details with ELF (PR #77010)

[Martin Storsjö via llvm-commits]

mstorsjo wrote:

> The patch looks good to me, but should it have a test case? I guess it doesn't matter much in this case, but it also seems easy enough to have one.

Not sure we explicitly want to have a test to disallow specific spellings of all potential options (unless we have a real world case for that) - the main reason here is that when looking into adding more options from lld/ELF into lld/MinGW, I noticed this discrepancy - which seems unintentional (all uses of the option use this one single form).

https://github.com/llvm/llvm-project/pull/77010