Update 2D uncertainties and intrinsic variance exercises

[jobovy]

This PR fixes the code for the exercises with uncertainties in x and y and updates those plots (see #18).

I also worked out how to fix the intrinsic variance objective function (which is also wrong in the paper currently I think). I think it should be

ln L = -0.5 \sum ( Delta_i^2 / [Sigma_i^2 + V] + ln[ Sigma_i^2 + V] + ln[ 1+m^2] )

so there's an additional term again. I think this agrees with @dfm's solution here (only saw this after I worked mine out, so independent).

I think I've coded that likelihood up correctly, but am getting strange results (see ex16.pdf and ex17.pdf in the updated files). @dfm, have you tried your code on the HBL data?

P.S.: the data was missing in the src/ directory, so I have also added files with the mock data, so the code should run easily now.

This PR should probably only be merged once @davidwhogg also updates the document.

[davidwhogg]

@jobovy Do you want to take a shot at updating the words and equations in the relevant part of the text, or are you suggesting / requiring that I do that? :)

[jobovy]

I was suggesting you should do it 😄

[davidwhogg]

Okay I made text updates with which I am not perfectly happy. I am merging this because we have something conceivable now.