[daip] calib solution intervals

[Eric Greisen]

Andy Biggs wrote:

Frazer Owen has made a practical suggestion - the task already insures 
that the last integration, if it is very near the last time, includes 
the last time.  We could change that to e.g. 50% of SOLINT near so that 
the last integration could be up to 50% longer than the usual, but that 
would keep the last one from being less than 50% of usual.  One can 
debate the 50% value - perhaps 70% or 60%.  This is easy and lets one 
get away with ignoring the deeper, not solvable, questions.

Eric