Prepare 1.12.0

[jgonggrijp]

The main reason for issuing a release now is the bugfix in #2876, but I'm bumping the minor version number rather than the patch number, because the new feature from #2877 already slipped into master before that. I'm not planning to draw special attention to the latter at this time, because the corresponding update to Contrib hasn't happened yet.

@jashkenas I have already updated all the generated files. Only the following steps remain for you, once you've approved this PR (edit: the only interesting part of the diff is the index.html, which contains the new change log entry and a new section about engine compatibility):

• Replace the TODO inline comment on lines 2656-2658 of the index.html by the date of release.
• Merge prepare-1.12.0 into master.
• npm publish: this will re-generate the bundles and annotated sources again, but should leave your working tree in a clean state. You'll get that warning again about mixing named and default exports, possibly twice, which you can still ignore. It will disappear in version 2.0.
• Create and push the 1.12.0 tag.
• Merge master into gh-pages.

@sktguha This update will not include your planned optimization of _.isEqual yet, but rest assured that your contribution is still welcome.

@joshuacc FYI: _.toPath will be in this release.

[coveralls]

Coverage remained the same at 96.305% when pulling b47fc2f on prepare-1.12.0 into d9741b3 on master.

[coveralls]

Coverage remained the same at 95.175% when pulling 8274127 on prepare-1.12.0 into ee99923 on master.

[jgonggrijp]

Regarding the failing build: that's SauceLabs failing to spin up outdated browsers again. I'll create an issue for that. The tests all pass, as you can see by scrolling upwards from this line.

[jgonggrijp]

Hold on, I started fiddling with the Sauce labs tests and then found out that there are still some minor compatibility issues with _.isEqual. I'll fix those first and then re-request a review.

[jgonggrijp]

Alright, fixed some more bugs and compatibility issues, kicked up some performance and finally got the Sauce tests to run smoothly most of the time. Really ready for review now.

(Travis ran twice because this is both a PR and a branch on the main repo. The second build failed because Coveralls can't handle checking the same commit twice. I'll avoid this situation in the future by only issueing PRs from my fork.)