Make fake Sherlock

[gpfrancis]

Make a fake version of Sherlock that returns random (but deterministic) classifications.

[RoyWilliams]

• Each classification AGN, BS, CV, NT, ORPHAN, SN, VS should be assigned at random with probability 1/7
• There will be no UNCLEAR
• If the classification is SN or NT, there should also be a distance and/orz and/or photoz in the sherlock packet each with probability 1/2, or else zero.
• Distance (Mpc) should be chosen as math.exp(random.uniform(2.5,7)), which gives a range 12 to 1100, biased to the smaller values
• The z and photoz can each be computed as 0.00023 * math.exp(random.uniform(2.5,7))
• Example: classification=SN, distance = exp(3) = 20.08 Mpc, z = 0.00023*exp(6) = 0.093, photoz = 0

[RoyWilliams]

The RA and Dec of the alert can be randomised at the same stage by the fake Sherlock. These sprinkle the alerts over the whole sphere

RA   = random.uniform(0.0, 359.999)
Dec = math.asin(random.uniform(-1, 1)) * 180/math.pi

[gpfrancis]

Merged into develop.