* significant improvement in microformats markup

This is a huge improvement and I believe will close #965.

Had these microformats reviewed by others in the community to help
ensure they are at least correct, if not complete.

I did not want to change the structure of the page, and so there it does
not fully mark up the entire ancestry chain, or reply chain, only the
direct decendants and direct ancestors are correctly associated, but
this is likely fine as the most important bit is to have access to the
urls for those toots which are now correctly fetchable.

* improve code climate

* trying to pass code climate tests

* code climate

* fix p-summary for content warning posts

* fix error introduced when merging via github

* Remove trailing whitespace

* Use query methods instead of explicit .blank? checks

* Add WebfingerResource class to extract usernames

* Use WebfingerResource in xrd#webfinger

The `Status` class has a default order on it, so when this query gets built and
gets all the way to `find_in_batches` there is an order already there.

When `find_in_batches` is run it discards any existing order on the query, and
emits a warning to the logs if there is one there.

This change removes the order prior calling `find_in_batches`, which will stop
the logged warning from occurring as well.

* Consolidate webfinger string creation under Account#to_webfinger_s

* Introduce Account#local_username_and_domain for consolidation

Checking reblog vs original status was happening in multiple places
across the app. For views, this logic was encapsulated in a helper
method named `proper_status` but in the other layers of the app, the
logic was duplicated.

Because the logic is used at all layers of the app, we extracted it into
a `Status#proper` method on the model and changed all uses of the logic
to use this method. There is now a single source of truth for this
condition.

We added test coverage to untested methods that got refactored.

* Rewrite Atom generation from stream entries to use Ox instead of Nokogiri::Builder

StreamEntry is now limited to only statuses, which allows some optimization. Removed
extra queries on AccountsController#show. AtomSerializer instead of AtomBuilderHelper
used in AccountsController#show, StreamEntriesController#show, StreamEntryRenderer
and PubSubHubbub::DistributionWorker

PubSubHubbub::DistributionWorker moves n+1 DomainBlock query to PubSubHubbub::DeliveryWorker
instead.

All Salmon slaps that aren't based on StreamEntry still use AtomBuilderHelper and Nokogiri

* All Salmon slaps now use Ox instead of Nokogiri. No touch from status on account

This adds the ability to filter user signup with a whitelist
instead of or in addition to a blacklist.

Fixes #817

make regeneration worker unique, (only schedule/execute once at a time)

do not re-create a dynamic class on each feed push call, make sure redis-rb uses hiredis

adjusted to allow choosing it, yet

reinsert it when someone boosts it

status is not in the system at time of distribution; fix #607 - reset privacy
settings to defaults when cancelling replies

redis-backed "mentions" timeline as redundant (given notifications)

fix ProcessFeedService pushing status into distribution if called a second time
while the first is still running (i.e. when a PuSH comes after a Salmon slap),
fix not running escape on spoiler text before emojify

Federate spoiler_text using warning attribute on <content /> instead of a <category term="spoiler" />
Clean up schema file from accidental development migrations

affected, replaced onClick handler with onMouseUp/Down to detect text
selection not trigger onClick handler then

background as well since it's a computationally expensive

now return empty hash instead of throwing a 404 error. When following,
merge into timeline will filter statuses