Default number of bins in Phi average is not ideal

[Adrian Rennie]

20 is a poor default choice as it gives 18 degree bins. Something that was divisible by 4 and 6 would be better (most likely symmetries). Thus 36 would probably be good, but that may be too fine a division for most data. 18 is a good compromise - 20 degree bins including bins centered on 0, 90, 180 and 270 degrees.

[gonzalezm]

Done! Default number of bins is 36 now and the first bin is defined to be centred at 0. File ring_testdata.txt changed correspondingly in order to pass the unit test.