This was previously mostly working because the allocator itself was
moving forward, but since each radio has its own allocator, aircraft
with different radios would often get overlapping intra-flight
frequencies.
Adds the following:
* AJS37
* AV-8B
* JF-17
This does move the preset channel allocation logic into its own class,
since we need to customize that behavior for the AJS37 since it has a
rather unique preset channel layout (see the comments in
`ViggenRadioChannelAllocator` for details).
Since we create a target waypoint for every target in a
strike/SEAD/DEAD objective area (including every ground vehicle), the
kneeboard can quickly be overrun with target waypoints. When there are
many target waypoints, collapse them all into a single row for
brevity.
Make the type of the waypoint a non-optional part of the constructor.
Every waypoint needs a type, and there's no good default (the previous
default, `TAKEOFF`, is actually unused). All of the target waypoints
were mistakenly being set as `TAKEOFF`, so I've fixed that in the
process.
Also, fix the bug where only the last custom target of a SEAD
objective was being added to the waypoint list because the append was
scoped incorrectly.
The first waypoint is automatically added by pydcs, so it's not
actually in our waypoint list from the flight planner. Import is from
the group so it shows up in the kneeboard.
Not every aircraft has a pydcs radio index, so we can't use that to
index into a list. Any mission with an A-10C crashes, since it would
try to use `None - 1` to index into the list of radios to find the
intra-flight radio.
Also fix the radio ranges for the newly added radios. The current
implementation can't model gaps, so extending the radio ranges across
those gaps means that we might allocate channels that aren't tunable
by those radios. Additionally, the end frequency is exclusive rather
than inclusive, so fix the ranges to include that last tunable
frequency.
I've been wrongly importing these from `pydcs.dcs` instead of just
`dcs`, because that was what PyCharm thought they were. These will all
be broken when we get back to using a real pydcs instead of relying on
its directory being in our tree.
This page in the wiki should be updated:
https://github.com/Khopa/dcs_liberation/wiki/Developer's-Guide
Instead of recommending that `PYTHONPATH` be updated in the run
configuration, it should instead recommend that Settings -> Project:
dcs_liberation -> Project Structure be set to exclude the pydcs
directory from the dcs_liberation content root, and add the pydcs
directory as a *separate* content root.
Alternatively, we could recommend that configure a virtualenv (good
advice anyway, and pycharm knows how to set them up) that have people
run `pip install -e pydcs`.
I think even easier would be switching from the virtualenv-style
requirements.txt to pipenv, which can actually encode the `-e` style
pip install into its equivalent of requirements.txt.
