adjust_size_for_badblocks reduces the operation size to account
for the block skipping done by the read/write functions when an
interval (partition name or whole chip) is specified rather than a data
amount.

Erase does not do block skipping, except for erase.spread which takes
a data amount rather than an interval (and thus already does not call
adjust_size_for_badblocks).  Calling adjust_size_for_badblocks when
block skipping is not done means that if bad blocks are present,
the "nand erase.part" and "nand erase.chip" commands will fail to erase
blocks at the end of the interval.

Signed-off-by: Scott Wood <scottwood@freescale.com>
Cc: Harvey Chapman <hchapman@3gfp.com>
Acked-by: Heiko Schocher <hs@denx.de>