Expand test cases for sequence interpolator

[reneburghardt]

From Polly:
a)

add a test case when start time is non-zero (might cause numerics issues)

b)

add irregular option for timestamps

c)

where the phase shifts are above the target sampling rate (now they are all smaller)
PS for clarity - target sampling rate is the one we resample data later, not the one we generate it with

d)

we need more generations for phase_shifts as the whole point is that they change the indexes of the data returned (eg for neuron 1 we return indexes 0 to 10 and for neuron 2 we return indexes from 1 to 11 because neuron 1 has a bigger timeshift than neuron 2)

e)

where the expected would be computed with np.interp (maybe removed functions from utils might help)
this would basically test the assumptions on how we do it

see #74

[pollytur]

more comments from that PR

• could you please add for the future backlog - to create the tests where the phase shifts are above the target sampling rate (now they are all smaller)
  PS for clarity - target sampling rate is the one we resample data later, not the one we generate it with