[MTD] DataFlash: bugfix, binary page sizes now handled

The wrong version of the "teach dataflash about binary density" patch
just got merged (v2 not v3) ... this restores the missing updates:

  * Fix the cmdlinepart *regression* that caused testing failures (!!)
    by restoring the original part labels in relevant cases.

  * Don't reference things that don't exist (!)
	- An opcode that doesn't even exist for DataFlash
	- The part is "at45db642" not "at45db641"
	- ID zero in this JEDEC table

  * Make the JEDEC probe routine report and handle errors better:
	- If the SPI calls fail, return the error codes.
	- Don't depend on ordering of table entries.
	- Unrecognized ids are different from parts that have no ID.
          We won't actually know how to handle them correctly; display
	  the ID and ignore the chip.

  * Move the original block comment about the "legacy" chip ID scheme
    back next to the code to which it applies ... not next to the new
    JEDEC query code, which uses an entirely different strategy.

  * Don't print a guessed erasesize; /proc/mtd has the real value.

And add a few more comments.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Cc: Bryan Wu <cooloney@kernel.org>
Cc: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>