[Edserplo_users] Another update: more Edteva-type stuff

[Steve Loch]

I've corrected the interpolation code used to fill gaps and rewritten the inner loop for the SeriesTimPage (items (a) and (e) from previous e-mail).

The former was a showstopper (no workaround and totally wrong). However there are caveats:-

1) The messages for incorrect endpoints need to be revised somewhat.

2) The interpolated values don't mirror Edteva's particularly well - i.e. we're not using the same algorithm. Edteva uses NAG's  E01BAF for spline calculation and E02BFF for evaluation -  but these are not available in Java so I need to investigate, starting with the NAG documentation. Previous research on this topic (by Samrat) indicated the match was pretty good except at the ends. The latest results seem to indicate that there are more serious reservations. Having identical algorithms with spot-on replication makes life much easier when comparing Edserplo and Edteva.

I think the interpolation is good enough for the moment whilst we're in the transition phase. The interpolations can be re-done in Edteva.

Steve