time-1.7 counts rusage wrong on Linux

time-1.7 converts rusage values from page size units to K, unconditionally,
but this is wrong at least on Linux, where they are already reported in K.
That results in time printing out values that are four times too big.

The source contains the comment "Note: Some machines express getrusage
statistics in terms of K, others in terms of pages." but has no actual code
that attempts to tell or ever works for those "some machines".

Thoughts?

Please keep me in the reply list on replies; I'm not subscribed to
bug-gnu-utils.

Thomas