Bump to v1.6.0, drop py3.8

[dachengx]

Bump minor version because we dropped py3.8

[coveralls]

coverage: 91.433% (+0.05%) from 91.388%
when pulling dda2f26 on bump_v156
into 882a8bf on master.

[WenzDaniel]

I think we had this numerical error before:

 aftobs = 3.469446951953614e-18, aft = 0.125, s1tot = 2.652351386976591

    @settings(deadline=None)
    @given(strategies.floats(0.0, 1.0), strategies.floats(0.01, 0.99), strategies.floats(2.0, 1000))
    def test_patternfit_stats(aftobs, aft, s1tot):
        s1top = aftobs * s1tot
>       assert (binom_test(s1top, s1tot, aft) >= 0) & (binom_test(s1top, s1tot, aft) <= 1)
E       assert (1.0000000000000002 >= 0 & 1.0000000000000002 <= 1)
E        +  where 1.0000000000000002 = binom_test(9.202192435055875e-18, 2.652351386976591, 0.125)
E        +  and   1.0000000000000002 = binom_test(9.202192435055875e-18, 2.652351386976591, 0.125)
E       Falsifying example: test_patternfit_stats(
E           aftobs=3.469446951953614e-18,
E           aft=0.125,
E           s1tot=2.652351386976591,
E       )
E       
E       You can reproduce this example by temporarily adding @reproduce_failure('6.93.1', b'AXicY2BgODyBAQYOKMCZDSKcnCyMDE8BPbIDxA==') as a decorator on your test case

We should try to address it and make the requirements maybe less tight. However, no show stopper.