Interp1() usability bug

[fork]

(I am not sure if this is worth a bug report, or what the ML behavior is, but it
might be worth considering.)

(1) interp1 will quietly allow different lengths for x and y inputs.  This is
alright, I guess, but shouldn't we throw a warning at least?  For example

14> interp1((0:5), 2*(0:5), 4.5)
ans =  9
15> interp1((0:6), 2*(0:5), 4.5)
ans =  9
16> interp1((0:5), 2*(0:6), 4.5)
ans =  9

(2) This behavior becomes a problem when we try to extrapolate, giving us an
unexplained error from inside the file.  The first three lines below are
expected, but the last is funky.

19> interp1((0:10), 2*(0:10), 4.5)
ans =  9
20> interp1((0:10), 2*(0:10), 11)
ans = NA
21> interp1((0:10), 2*(0:10), 11, 'extrap')
ans =  22
22> interp1((0:10), 2*(0:9), 11, 'extrap')
error: A(I): Index exceeds matrix dimension.
error: called from:
error:   C:\Octave\3.2.4_gcc-4.4.0\share\octave\3.2.4\m\general\interp1.m at
line 208, column 19

If we checked for length(x) == length(y) and errored out if not, we could avoid
this.  Otherwise, shouldn't we at least give a slightly informative error,
rather than just letting the malfunctioning indexing break it for us?

I can file a bug tonight if its worth it.