Remove official Appveyor support (#2386)

docs/working-examples.md

@@ -75,7 +75,7 @@ title: Working examples
 | [KDEpy][]                         | ![github icon][] | ![windows icon][] ![apple icon][] ![linux icon][] | Kernel Density Estimation in Python |
 | [dd-trace-py][]                   | ![github icon][] | ![windows icon][] ![apple icon][] ![linux icon][] | Uses custom alternate arch emulation on GitHub |
 | [tgcalls][]                       | ![github icon][] | ![apple icon][] ![windows icon][] | Python `pybind11` binding to Telegram's WebRTC library with third party dependencies like `OpenSSL`, `MozJPEG`, `FFmpeg`, etc. |
-| [python-rapidjson][]              | ![travisci icon][] ![gitlab icon][] ![appveyor icon][] | ![windows icon][] ![linux icon][] | Python wrapper around rapidjson |
+| [python-rapidjson][]              | ![travisci icon][] ![gitlab icon][] | ![windows icon][] ![linux icon][] | Python wrapper around rapidjson |
 | [pybind11 python_example][]       | ![github icon][] | ![windows icon][] ![apple icon][] ![linux icon][] | Example pybind11 module built with a Python-based build system |
 | [sourmash][]                      | ![github icon][] | ![apple icon][] ![linux icon][] ![windows icon][] | Quickly search, compare, and analyze genomic and metagenomic data sets. |
 | [abess][]                         | ![github icon][] | ![windows icon][] ![apple icon][] ![linux icon][] | A fast best-subset selection library. It uses cibuildwheel to build a large project with C++ extensions. |
@@ -115,7 +115,7 @@ title: Working examples
 | [CorrectionLib][]                 | ![github icon][] | ![apple icon][] ![linux icon][] | Structured JSON powered correction library for HEP, designed for the CMS experiment at CERN. |
 | [xmlstarlet][]                    | ![github icon][] | ![windows icon][] ![apple icon][] ![linux icon][] | Python 3.6+ CFFI bindings with true MSVC build. |
 | [werpy][]                         | ![github icon][] | ![windows icon][] ![linux icon][] ![apple icon][] | An ultra-fast python package using optimized dynamic programming to compute the Word Error Rate (WER). |
-| [pyinstrument_cext][]             | ![travisci icon][] ![appveyor icon][] | ![windows icon][] ![apple icon][] ![linux icon][] | A simple C extension, without external dependencies |
+| [pyinstrument_cext][]             | ![travisci icon][] | ![windows icon][] ![apple icon][] ![linux icon][] | A simple C extension, without external dependencies |
 | [pybind11 cross build example][]  | ![github icon][] ![gitlab icon][] | ![windows icon][] ![apple icon][] ![linux icon][] | Same as pybind11 cmake_example but used to demo Linux ARM + Windows + macOS builds on GitLab |
 
 [scikit-learn]: https://github.com/scikit-learn/scikit-learn
@@ -228,7 +228,6 @@ title: Working examples
 [pyinstrument_cext]: https://github.com/joerick/pyinstrument_cext
 [pybind11 cross build example]: https://github.com/wbarnha/pybind_cmake_example_crossbuild
 
-[appveyor icon]: data/readme_icons/appveyor.svg
 [github icon]: data/readme_icons/github.svg
 [azurepipelines icon]: data/readme_icons/azurepipelines.svg
 [circleci icon]: data/readme_icons/circleci.svg