Use chrome data when android is absent

[JLHwung]

Q	A
Fixed Issues?	Fixes #10789
Patch: Bug Fix?	Yes
Tests Added + Pass?	Yes
License	MIT

In this PR we introduce a getLowestImplementedVersion which adds chrome support data as fallback when android support data is absent, by doing preset-env would not introduce unnecessary transforms for Android >= 52, which should share the same features sets as Chrome >= 52.

The previous behaviour is revealed in this CI failure. This behaviours renders settings like last 2 versions unusable and preset-env will thus apply all the transforms.

[zloirock]

In core-js it's implemented as https://github.com/zloirock/core-js/blob/v3.4.5/packages/core-js-compat/index.js#L41

[nicolo-ribaudo]

Maybe this should be added to https://github.com/browserslist/browserslist/blob/4da6f9c683ce27816f8804663c4cd29ef38d01a7/index.js#L454?

[zloirock]

The existence of any android versions greater than 4.4.4 in browserslist looks like a bug.

[JLHwung]

According to the caniuse description, the android whose version are greater than 4.4.4 refers to Android 5-6.x WebView. We could propose to add a new android-webview agent to caniuse-db but I think a single android works fine here as long as the version does not overlap with Android Browser.

My first approach was somehow similar to core-js-compat: map android >= 36 to chrome >= 36. But I realize that it would make queries like android >= 4 converted to { android: "4.0.0", chrome: "76.0.0" } so I switch to current implementation.

@zloirock Wouldn't current approach in core-js-compat map android 5 to chrome 5? I think we should emit warnings for 5 <= android < 30 in which both WebView and Android Browser are not available.

[zloirock]

@JLHwung browserslist will throw an error on android 5, so not.

For Android 5+ Chrome and WebView we have and_chr -> chrome.

See compat-table/compat-table#801

[JLHwung]

The plugins listed here is generated from a sort-of modern browserslist configuration. It can help us prioritize bug fixes in the current transforms. because these plugins are expected to be used for a longer time than other standard transforms.

[JLHwung]

Recently the chrome 49 usage just drops to 0.5% and therefore excluded from browserslists defaults > 0.5% now.