Added Continuous Integration.

[tdulcet]

• Added Continuous Integration (CI).
  • Added both Travis CI and GitHub Actions CI. See the logs here and here respectively.
  • Need to enable at travis-ci.com before merging this pull request.
  • They will both automatically build GpuOwl on every commit and pull request. GitHub Actions CI will additionally build it every month in case there has been no commits.
  • Builds on Ubuntu 18.04 and 20.04 with both GCC and Clang.
  • Builds on Windows.
  • Runs the Cppcheck and ShellCheck linters.
• Enabled -O3 optimization in the Makefile.
• Updated the READMEs.
  • Noted that GpuOwl also works with Nvidia and Intel GPUs supporting OpenCL.
  • Noted that GpuOwl requires C++20 support (see note below)
  • Updated command line options section to reflect the current gpuowl -h output. Fixes #209 and #179.
  • Improved Markdown formatting.
  • Minor copy edits.

Notes:

• https://github.com/preda/gpuowl/commit/817a9831dc9fdc595709bd5f68482e654b758b65 broke Clang support, so the CI currently allows those builds to fail. This should probably also be noted on the README.
• GCC on Ubuntu 18.04 also fails because of some unsupported C++20 syntax beginning with https://github.com/preda/gpuowl/commit/efbbcbf4bc2272d941cbd442c9345e2fb6b5ab1d, so the CI currently also allows those builds to fail. GCC 7 only supports up to C++17.
• The Windows build on Travis CI is failing because the Python 3 binary is named python when installed via the Chocolatey package manager and the Makefile currently expects python3, so the CI currently allows this build to fail.
• The Windows build on GitHub Actions CI is failing because mingw-w64 is using GCC 8, which does not support std::filesystem on Windows, so the CI currently also allows this build to fail. See bug 78870.

See the logs for more information on these issues. Currently only the GCC on Ubuntu 20.04 builds are not allowed to fail with the CI, although this can be changed if/when the above issues are fixed.

@Danc2050 and I are considering updating our GPU notebook for Google Colaboratory to use GpuOwl instead of CUDALucas, so we need latest version to always successfully build. This PR would allow users to see directly on the top of the README if the latest version of GpuOwl builds by checking the badges.

[preda]

Thank you!

[sethtroisi]

The README updates are very nice. Thanks!