hoon: minor improvement to +mod and +dvr

[ashelkovnykov]

Unify logic for div and mod in dvr, instead of splitting the logic and repeating the same calculation twice on a dvr call.

Also add default sample to dvr, mirroring div and mod.

Tested changes successfully against latest code from develop (both urbit/urbit and urbit/vere).

[joemfb]

With this change, the +mod jet should also be updated to use u3m_error() as the other two do.

[ashelkovnykov]

@joemfb Good catch - added a companion PR to urbit/vere: #257. They must be released together.

[philipcmonk]

Is this a semantic change? Is it possible that the old code would give a different result than the new code? If so, it must be released in a hoon kelvin update so the jet never fires with the wrong nock

[ashelkovnykov]

@philipcmonk The only difference should be the one Joe pointed out: the new code gives a different result in the case of "divide by zero" error.

Old code = bail: exit
New code = u3t_mean("divide-by-zero"), then bail: exit

What's the protocol for explicitly marking this PR as requiring a hoon kelvin update? Should I dump the kelvin version in this PR, and then all other such changes nicely merge together since they also dump the kelvin version?

[joemfb]

Will it's technically true that this code must be released at the same time as the paired jet changes to avoid a mismatch, it's an utterly trivial mismatch (the error string in the trace when dividing by zero). The livenet version of this code already contains such a mismatch -- the +dvr jet includes the error string, while the hoon does not. And stack traces are already subject to such mismatches anytime the set of actual jets changes relative to execution history (ie, on replay) -- a problem for another day. I don't think we have to worry about kelvins for this PR specifically.

this will do

[ashelkovnykov]

@belisarius222 @jalehman Ready for merge