benchmarks: Minor cleanup for strtr benchmark

benchmarks: Minor cleanup for strtr benchmark

Only bench with variable input ("indirect"). The static string is
too unrealistic and not worth testing for imho, if it even makes
a difference at all.

Also bump the default count from 100 to 10,000 given it's so tiny.

Change-Id: Iccc35af4bd3c5b0967983ddfacd1d4ead235e4a4