This diff does away with addEntriesForFunctionsWithoutUnwindInfo(),
because addSymbol() can now determine which functions need those
entries.
While overhauling UnwindInfoSection, I also parallelized the relocation
of the contents of the CUEs. This somewhat offsets the time regression
from creating one InputSection per CUE (which was done in D109944).
Please include the explanation on why we're doing this (for eg., "because we'll use it with parallelForEachN" )
(or perhaps move this comment to the symbols's decl statement.)