Drop support for float128 outside of linux and mac (#3930)

* Drop support for float128 outside of linux and mac

* Improve release notes

Author: lucianopaz

RELEASE-NOTES.md

@@ -28,6 +28,7 @@
   pointwise log likelihood (see [#3883](https://github.com/pymc-devs/pymc3/pull/3883)).
 - The multiprocessing start method on MacOS is now set to "forkserver", to avoid crashes (see issue [#3849](https://github.com/pymc-devs/pymc3/issues/3849), solved by [#3919](https://github.com/pymc-devs/pymc3/pull/3919)).
 - Forced the `Beta` distribution's `random` method to generate samples that are in the open interval $(0, 1)$, i.e. no value can be equal to zero or equal to one (issue [#3898](https://github.com/pymc-devs/pymc3/issues/3898) fixed by [#3924](https://github.com/pymc-devs/pymc3/pull/3924)).
+- Fixed an issue that happened on Windows, that was introduced by the clipped beta distribution rvs function ([#3924](https://github.com/pymc-devs/pymc3/pull/3924)). Windows does not support the `float128` dtype, but we had assumed that it had to be available. The solution was to only support `float128` on Linux and Darwin systems (see issue [#3929](https://github.com/pymc-devs/pymc3/issues/3849) fixed by [#3930](https://github.com/pymc-devs/pymc3/pull/3930)).
 
 ### Deprecations
 - Remove `sample_ppc` and `sample_ppc_w` that were deprecated in 3.6.

pymc3/distributions/dist_math.py

@@ -17,6 +17,7 @@
 
 @author: johnsalvatier
 '''
+import platform
 import numpy as np
 import scipy.linalg
 import scipy.stats
@@ -36,8 +37,13 @@
 c = - .5 * np.log(2. * np.pi)
 _beta_clip_values = {
     dtype: (np.nextafter(0, 1, dtype=dtype), np.nextafter(1, 0, dtype=dtype))
-    for dtype in ["float16", "float32", "float64", "float128"]
+    for dtype in ["float16", "float32", "float64"]
 }
+if platform.system() in ["Linux", "Darwin"]:
+    _beta_clip_values["float128"] = (
+        np.nextafter(0, 1, dtype="float128"),
+        np.nextafter(1, 0, dtype="float128")
+    )