Make load times faster for Chrome?

This came up in our meeting this morning.

In general, Firefox seems to have faster load times for large trees. See [this SO thread](https://stackoverflow.com/questions/59308105/why-is-firefox-much-faster-than-chrome-when-loading-big-documents/59438847#59438847) for details on why this may be the case (thanks @kwcantrell for finding this!).

It may be possible to get around this by forcing Chrome to defer loading large pieces of data (see that SO thread for one implementation), but as @ElDeveloper pointed out this would break Empress when looking at local HTML documents -- see [here](https://stackoverflow.com/a/11063963/10730311) for some background as to why this would anger browsers.

It may be possible to get around _that_ by only supporting using Empress through view.qiime2.org and through starting up a local python server, but in any case it'd still break `qiime tools view`. (We could handle this semi-gracefully by detecting if XMLHttpRequests cause an error and if so showing a message explaining to the user that only certain ways of using Empress are supported.)

... TLDR, less crazy performance fixes are likely a better idea for the short term :)

No files or tests are named. Reproduce loading a large tree in Chrome and Firefox through view.qiime2.org, qiime tools view, and a local Python server, then profile the slow path; done means faster Chrome loads without breaking local HTML usage or supported viewing workflows.

Written by the indexing model from the issue text.