Make wind speed moddable.

These should probably be overridable per theater and per season, but
even with that we'll want some defaults.

https://github.com/dcs-liberation/dcs_liberation/issues/2862

docs/modding/index.rst

@@ -6,4 +6,5 @@ Modding guide
    :caption: Contents:
 
    fuel-consumption-measurement.md
-   layouts.rst
+   layouts.rst
+   weather.rst

docs/modding/weather.rst

@@ -0,0 +1,76 @@
+#######
+Weather
+#######
+
+Weather conditions in DCS Liberation are randomly generated at the start of each
+turn. Some of the inputs to that generator (more to come) can be controlled via
+the config files in ``resources/weather``.
+
+**********
+Archetypes
+**********
+
+A weather archetype defines the the conditions for a style of weather, such as
+"clear", or "raining". There are currently four archetypes:
+
+1. clear
+2. cloudy
+3. raining
+4. thunderstorm
+
+The odds of each archetype appearing in each season are defined in the theater
+yaml files (``resources/theaters/*/info.yaml``).
+
+.. literalinclude:: ../../resources/weather/archetypes/clear.yaml
+  :language: yaml
+  :linenos:
+  :caption: resources/weather/archetypes/clear.yaml
+
+Wind speeds
+===========
+
+DCS missions define wind with a speed and heading at each of three altitudes:
+
+1. MSL
+2. 2000 meters
+3. 8000 meters
+
+Blending between each altitude band is done in a manner defined by DCS.
+
+Liberation randomly generates a direction for the wind at MSL, and each other
+altitude band will have wind within +/- 90 degrees of that heading.
+
+Wind speeds can be modded by altering the ``speed`` dict in the archetype yaml.
+The only random distribution currently supported is the Weibull distribution, so
+all archetypes currently use:
+
+.. code:: yaml
+
+   speed:
+     weibull:
+       ...
+
+The Weibull distribution has two parameters: a shape and a scale.
+
+The scale is simplest to understand. 63.2% of all outcomes of the distribution
+are below the scale parameter.
+
+The shape controls where the peak of the distribution is. See the examples in
+the links below for illustrations and guidelines, but generally speaking low
+values (between 1 and 2.6) will cause low speeds to be more common, medium
+values (around 3) will be fairly evenly distributed around the median, and high
+values (greater than 3.7) will cause high speeds to be more common. As wind
+speeds tend to be higher at higher altitudes and fairly slow close to the
+ground, you typically want a low value for MSL, a medium value for 2000m, and a
+high value for 8000m.
+
+For examples, see https://statisticsbyjim.com/probability/weibull-distribution/.
+To experiment with different inputs, use Wolfram Alpha, e.g.
+https://www.wolframalpha.com/input?i=weibull+distribution+1.5+5.
+
+When generating wind speeds, each subsequent altitude band will have the lower
+band's speed added to its scale parameter. That is, for the example above, the
+actual scale parameter of ``at_2000m`` will be ``20 + wind speed at MSL``, and
+the scale parameter of ``at_8000m`` will be ``20 + wind speed at 2000m``. This
+is to ensure that a generally windy day (high wind speed at MSL) will create
+similarly high winds at higher altitudes and vice versa.