Better content type detection for urls.

[vangorra]

Uses the content type supplied by the server when

[dbashford]

Changing the signature for fromFileWithPath will likely break the tests and would at the very least require doc updates. And I rather not complicate this.

Can you instead just add a typeOverride to the options object and use that?

[vangorra]

Good call! Here you go.

[dbashford]

Could you give me an idea of a file where the mime type wasn't being figured out properly? I can build a test around that later.

[vangorra]

Without this PR, mime.lookup uses the aspx extension to resolve an octet stream as the mimetype. Which textract will not parse. The following URL will demonstrate this.
http://apps.leg.wa.gov/billinfo/summary.aspx?bill=1276

[dbashford]

👍

I'm deeply hoping to get another chunk of work done towards 2.0 this weekend. I intend to merge this, but I want to do it when I'm ready to write a test or two for it. So bear with me. =)