Use Chi^2/KS for mechanical 0-D/1-D simulator-assessor agreement testing for all Lite SPs of appropriate type

[axch]

• One-sample tests in both cases
• The K-S test would need to do (or accept the results of) quadrature on the alleged density function (which it might also normalize) to compute the CDF at all needful points
• Bounds of integration should be extractable from the type, up to occasional exceptions
• Only applies to SPs whose types indicate that their output is discrete or 1-D
• Only directly applies to Lite SPs, but we can test method-wise agreement with Puma separately (the latter ideally with enough compute to reliably complain about the beta clamping bug, which we can then fix)

[axch]

Relevant code locations:

• Existing randomized tests in test/properties
• Perhaps the Venture Lite type system in types.py
• The QuickCheck like property testing system in test/randomized.py and test/random_values.py, which probably doesn't need to be changed but is likely to be useful.

[axch]

Much though I like the algorithmic challenge of doing a K-S test for agreement between a sampler and a probability density function by running the quadrature needed to compute the cdf from the pdf in tandem with the cdf queries the K-S test requires and cleverly reusing query points, probably the cleanest way to achieve the desired effect for 1-D continuous distributions would be to just reuse analytically known cdfs.

To wit:

• define a new Lite PSP method named "cdf" (possibly also "sf" for the survivor function, if desired)
• add cdfs for the SPs in Lite
• add mechanical tests that numerically differentiate the cdf and compare to the pdf
• add mechanical tests that do K-S of a sample drawn by the simulator against the cdf

[axch]

Finite-domain discrete distributions are easier, because the exact answer for the Chi^2- or G-test can be computed by enumerating all values, because the pdf is also the probability mass function.