If there are NaNs in the posterior, continue onward but print a message

See merge request !312

Resolve "Discrepancy in definitions of matched filter SNR"

Closes #255

See merge request !306

update with catalog events

See merge request !307

NaNs in the posterior are generted by the conversion function (say when
the prior is fixed to zero for the spins). This causes the
plot_marginals to fail. Instead, a message will be printed, but the
script will otherwise continue.

add conversions documentation

See merge request !308

fix ordering after interpolation

See merge request !301

Resolve "Prior read in from file chokes on empty lines"

Closes #265

See merge request !303

Resolve "copying BBHPriorDict"

Closes #258

See merge request !305

Resolve "Calculate time_delay_from_geocenter with coalescence time"

Closes #264

See merge request !304

Fixing some argument bugs in pymc3.py.

See merge request !299

Further fix to encapsulation functions

See merge request !297

- Fixes bug in which the arguments here not converted to floats
- Fixes typo in lamda1 -> lambda2

allow aligned spin (chi) for BBH and non-aligned spin for BNS

See merge request !292

make plot marginals work

See merge request !294

Added tutorial for fitting a line with x and y

See merge request !284

pymc3.py: explicitly set step method arguments

See merge request !293

 - PyMC3 only applies the step_kwargs and nuts_kwargs arguments
   to the step methods if they are automaticallu allocated. Otherwise,
   it assumes you have already initialised the method with the
   required keywork arguments. This was not being done in bilby, so
   this patch fixes it and does explicitly initialise the step methods
   with values passed to step_kwargs or nuts_kwargs as required.

Attempt to allow compound step methods for PyMC3

See merge request !289