Set GC threshold when mounting with checkpointing

This steps up our garbage collection threshold over time when mounting
userdata with checkpointing=disable for f2fs. With this scheme, we step
up the percent of disk we will tolerate being unusable during
checkpointing. At 100%, the filesystem will be able to always mount.
This means we will attempt to mount at most 10 times, for a max time of
about 50 seconds.

Also logs mount time under ro.boottime.init.mount.[target]

If the device does not support checkpoint=disable:[n%], it will mount
with checkpoint=disable, which is equivalent to checkpoint=disable:0%

Test: Boot in checkpointing mode
Bug: 150506496
Change-Id: I047ff98513f4ea832e1216b91e454a57fd2e8bf6