[PATCH 1 of 4 mergedriver] debugmergestate: explain why we create mergestate objects directly

[Siddharth Agarwal]

On 11/17/15 22:56, Martin von Zweigbergk wrote:
> This explains why it's okay to do it, but why is it not okay to call 
> read() instead? I guess read() is the alternative here that you want 
> to avoid. The comment in the patch body explains even less about the 
> reason for avoiding it. I don't mind fixing up in flight if you 
> provide a text

Yeah, sorry, meant to add that before sending but forgot.

How about this for text?

We would normally use the read() constructur, but in this case it's fine
because
- we implement our own reading layer, so the extra parsing done by
   read() is unnecessary
- read() can throw an exception for unsupported merge state records,
   but here we'd like to handle that separately
- debugmergestate needs to be privy to mergestate internals anyway