Provide an option to force synchronous initialization of inode blocks

[markj]

Summary

Per the comment, we use an ordered write to guarantee that the
initialized block is written before the updated cylinder group header.
However, I've found that a number of GEOMs don't properly handle
BIO_ORDERED, so babarrierwrite() doesn't necessarily work as advertised.
I plan to fix at least gmirror, but in the meantime, I believe it's
useful to have a flag to work around the problem. The default value
configures the current behaviour, i.e., inode initialization is still
done asynchronously.

In a test where I untar a large file hierarchy onto a newly initialized
filesystem, the time required goes up by about 50% with
doasyncinodeinit=0 configured.

Longer-term I wonder if it might be possible to integrate this
dependency into SU. It doesn't look like we handle the case where
an async inode initialization returns with a write error, but the CG
update succeeds.

[mckusick]

I believe that the correct solution is to make barrier writes work correctly as they can be used to solve a multitude of problems. Indeed I would use them more extensively in UFS if I believed that they worked. That said, I suggest a comment over the definition of the doasyncinodeinit that explains that synchronous inode initialization is only needed when barrier writes do not work.

[kib]

I agree with Kirk.