Embed bootstrapping into fitsignal

[luisfabib]

Internalize the bootstrapping of all fitsignal quantities to ease the use of bootstrapping for standard workflows. The user can change between covariance and bootstrapped uncertainty quantification via the a new option:

fit = dl.fitsignal(V,t,r,uq='covariance') # covariance-based uncertainty quantification
fit = dl.fitsignal(V,t,r,uq='bootstrap') # bootstrapped uncertainty quantification

By default, the bootstrap will be done with 1000 samples to ensure quality. Still, the number of samples can be specified as follows:

fit = dl.fitsignal(V,t,r,uq=['bootstrap',100]) # bootstrapped uncertainty quantification with 100 samples

Closes #75

Changelog:

• Major refactoring of fitsignal code
• Embed call of bootan into fitsignal
• Add new tests to check bootstrapped quantities
• Speedup fitsignal tests