[nfsv4] Some minor fixes to the pnfs-object-layout draft

[Boaz Harrosh <[e-mail]>]

I'd suggest two fixes to section 5.3. Data Mapping Schemes

1. Throughout the section there are Formulas with missing parenthesis
   for example:
	N = L / (W-P * stripe_unit)

   it is not nice that W-P is not inside parenthesis as proper in math annotation
   rules, even though the spacing does hint of the author intention in this case.

2. In: 5.4.4. PNFS_OSD_RAID_PQ
  It is made to believe that both a raid_4/fixed-parity-devices and raid5/rotate-parity-devices
  are possible. I think it is only rotate-parity-devices, right?

And if you are opening the draft for editing then we have that accumulated error extension
when out-of-memory condition happens at layout-return.
[ In short, a last error-component can have a NULL-out component descriptor with an error-value
  and an accumulated byte range. The byte range is the union of all failing ranges and the
  error-value is the most sever error-out of the lot, so for example one device had EIO and another
  ERESOURCE the EIO is returned. (with defined error severity)]

Thanks
Boaz