Packages

ios: pod 'oonimkall', :podspec => 'https://dl.bintray.com/ooni/ios/oonimkall-2020.11.12-200336.podspec'
android: implementation 'org.ooni:oonimkall:2020.11.12-200733'

Improvements

Explicitly add code of conduct based on Tor's code of conduct
experiment/urlgetter: add experimental support for http3 measurements
experiment/riseupvpn: add experimental support for measuring riseupvpn API endpoints
experiment/httphostheader: start sketching our experiment to check for Host-header-based blocking
experiment/tlstool: start exploring better TLS blocking measurements
geolocate: add more ways of finding the user's IP address
session: use cloudflare dns as the resolver since it's more reliable than powerdns.org

tests: We now run a smaller test suite at every commit, causing the coverage measurement to drop. This is actually an artefact caused by such a change in the measurement procedure. We run most other tests once per day nightly. When a branch is named ^release/*, however, we run much more stringent checks, because we want much more QA when we are about to bless a release.
dependencies: The Psiphon dependency is now vendored inside https://github.com/ooni/psiphon. This allows us to simplify dependencies management as well as to avoid requiring Go 1.15 users to specify -tags PSIPHON_DISABLE_QUIC thanks to small local changes.
tests: use https://ams-pg-test.ooni.org for integration tests

tests: make sure go test run -short ./... has a reasonably short runtime by flagging many integration tests as "long" and/or by rewriting long tests to become less dependent on external services and/or the network.
experiment/sniblocking: use example.org as test helper, since this seems more reliable than example.com
experiment/urlgetter: allow using a custom CA pool
libminiooni: support input files with CRLF line terminators on Unix systems
experiment.go: clarify when Run should return an error to the caller
oonimkall: emit progress events for tests taking input