pin update fixes

[Stebalien]

• flush pins
• correctly print output
• add a sharness test

[schomatis]

@Stebalien What issue/PR/doc would you recommend to learn about the pinning system architecture to better understand the Flush call?

[Stebalien]

Don't know. Looking at the code, it looks like pins are cached in-memory until the user calls Flush. At that point, they're packed into a single protonode and stored in the datastore. corerepo.Pin and corerepo.Unpin automatically flush (and have a bunch of other interesting logic) but pinner.{Pin,Unpin,Update} don't.

[schomatis]

There's some interesting logic down there..

[schomatis]

corerepo.Pin and corerepo.Unpin automatically flush (and have a bunch of other interesting logic) but pinner.{Pin,Unpin,Update} don't.

Sounds reasonable.

I can confirm this fixes some weird unwrapOutput error.

LGTM, minus the rebase, probably because I took to long to review this, sorry.

[whyrusleeping]

will merge in 0.4.18

[whyrusleeping]

@Stebalien needs a rebase, sorry.