Reasoning being:
* The result_set and other APIs are nicer to use.
* We use mariadb on our server rather than mysql, so this would minimize the chance of any incompatibilities.
* The mysql C++ connector 1.1 isn't compatible with with c++17 (https://stackoverflow.com/q/47284705).
Reasons:
* These projectfiles are still 32-bit, whereas all other builds for all other OSes are now 64-bit.
* Relatedly, these projectfiles are dependent on the libraries kept at the aquileia/external repository.
* VS2019 is still listed as supporting Windows 7, so there is a minimal likelihood of developers not being able to use VS2019.
* The VS2019 projectfiles instead use vcpkg to get the required dependencies, which is easier to setup than the aquileia/external prebuilt libraries.
* It's one less thing that needs to be updated whenever source files are added/moved/removed.
* It's two fewer jobs that Travis needs to run, which means Travis builds will finish more quickly.
The first run of the 2019 jobs will fail, since that run will be used to build and cache the vcpkg dependencies - there's no way to get enough time to build wesnoth and build the dependencies in a single job.
This also removes the separate WML_tests.cmd scripts from VC14 and VC16, now instead using the same run_wml_tests script as the linux jobs.
There is currently no caching of the Windows jobs, since msbuild unfortunately uses timestamps rather than more complete information like scons/ccache in order to determine if things need to be rebuilt. Since git doesn't preserve any sort of last modified timestamp, the cloned repo into the travis job is always "newer" than the cached compiled output, and therefore wesnoth is always fully rebuilt.
Ideally, assuming this sticks around and gets out of early access from travis, we could then use just travis instead of travis and appveyor.
