Use shared Markdown implementation

[gnprice]

For doing local echo when the user sends a message, the mobile app currently uses a version of the webapp's implementation, forked off in 2017. Predictably, the fork has not kept up with new features or bugfixes in the webapp.

Instead of a fork, we should share a single version. This would help bring the mobile app up to date with features like silent mentions, user and stream IDs in mentions, and @stream, as well as numerous bugfixes: I count at least 25 substantive changes since the fork in the webapp version of these files, out of 67 total commits that touch them.

The first stage of this will be to take the webapp's current implementation and move it to the @zulip/shared package found at static/shared/, refactoring as needed to make that possible. Once it's reusable and shared, it should be very smooth for the mobile app to switch to it -- the fork has very few changes of its own.

The forked code is found at https://github.com/zulip/zulip-markdown-parser, and is zulip-markdown-parser on NPM.

Many more details at zulip/zulip#13892, which is the web-side issue to move that code into the @zulip/shared package so that we can in turn use it here.

[gnprice]

@showell has recently done the work (in zulip/zulip#15898 and later changes) to make two of these files shared, namely fenced_code.js and emoji.js. So that makes this ready for us to start work on from our side.

(I think the code changes at this stage will actually happen in https://github.com/zulip/zulip-markdown-parser ; see detailed plan at the end of the description of zulip/zulip#13892 . But when we're ready to complete this issue, we'll cut that package out entirely and retire that repo.)