Suggestion for spike-field GLM on page /11.html

[abuzarmahmood]

It might be helpful to explain why the phase needs to be represented as sin(phi) and cos(phi) instead of regressing to phi directly using the GLM. It took me a bit to realize that log(lambda) cannot be linearly regressed to phase but sin(phi) and cos(phi) can.

Happy to create a PR if this seems useful.

[Mark-Kramer]

A very late reply ... correct, the discontinuity at 0/2pi makes regressing on phase problematic. Pull request very much appreciated!