No error thrown if invalid type provided, silently falls back to `scatter`

If you enter a type for a trace that is not supported in the package you have, Plotly simply falls back to scatter plot. This is incorrect, an error should be thrown to indicate the user entered an invalid type.

In my particular case, I had been using scattergl type with the full plotly.js-dist bundle, then when I switched our project over to plotly.js-finance (https://github.com/plotly/plotly.js/blob/master/dist/README.md#plotlyjs-finance) I thought everything was working correctly as all our test charts appeared correctly. However, I didn't notice plotly.js-finance did not include scattergl, so was in fact falling back to scatter, which had a significant performance penalty I did not notice while testing. I would have much appreciated an error indicating the type I was providing was not available in the partial bundle I was using.

I've made a code pen of the issue here, using a type "aninvalidtype": https://codepen.io/mofojed/pen/pojpVvP

Start with the linked CodePen reproduction using "aninvalidtype" and compare behavior between the full plotly.js-dist bundle and plotly.js-finance, especially for the unavailable scattergl type. Trace how the selected trace type is resolved in the package entry points. Done means unsupported types produce an error instead of silently rendering as scatter, with regression coverage for partial bundles.