Delicatessen

- Added Poisson regression model
- Added support for some survival analysis functionality (exponential, Weibull, Weibull-AFT models)

Adds functionality to detect `nan` values for the starting values. Since `nan` is not natively handled, this makes problems with estimating equations (returning `nan` or `nan` in the input data) easier to detect and correct.

Initial release of delicatessen

Fixing an error in the partial derivatives.

SciPy was updating the values of theta, which led to different answers. No substantive changes really occurred if `dx` was kept small, but large values of `dx` led to deviations. This has been patched, and a new test was added to check for this behavior.

Last planned beta release. Added custom, user-specified root-finding algorithms and additional options for numerical approximation of derivatives

Release consists of improved code comments and adds the `dx` option to control the spacing argument for the numerical derivative approximations