publish: should be able to identify removed remote clones and fail informatively

[yarikoptic]

(git)smaug:…btrfs/datasets/datalad/crawl/labs/poldrack/NP-180740/sourcedata[master]git
$> datalad publish --to=datalad-public -r --missing skip  # just let's do a full sweep
[INFO   ] Publishing <Dataset path=/mnt/btrfs/datasets/datalad/crawl/labs/poldrack/NP-180740/sourcedata> data to datalad-public 
Total (0 ok, 604 failed out of 604):   0%|                                                                                                                                                             | 0.00/10.4G [00:39<?, ?B/s]
[WARNING] Running copy resulted in stderr output: git-annex-shell: expected repository UUID d326b73e-1f9f-4ed0-97cb-1bbfd24f5057 but found UUID b38aead5-b9da-4fac-8588-286990fd2c94
rsync: connection unexpectedly closed (0 bytes received so far) [sender]
rsync error: error in rsync protocol data stream (code 12) at io.c(235) [sender=3.1.3]
rsync exited 12
git-annex-shell: expected repository UUID d326b73e-1f9f-4ed0-97cb-1bbfd24f5057 but found UUID b38aead5-b9da-4fac-8588-286990fd2c94
rsync: connection unexpectedly closed (0 bytes received so far) [sender]
rsync error: error in rsync protocol data stream (code 12) at io.c(235) [sender=3.1.3]
rsync exited 12
git-annex-shell: expected repository UUID d326b73e-1f9f-4ed0-97cb-1bbfd24f5057 but found UUID b38aead5-b9da-4fac-8588-286990fd2c94
rsync: connection unexpectedly closed (0 bytes received so far) [sender]
rsync error: error in rsync protocol data stream (code 12) at io.c(235) [sender=3.1.3]
rsync exited 12
git-annex-shell: expected repository UUID d326b73e-1f9f-4ed0-97cb-1bbfd24f5057 ... 
                                                                                                                                                                                                                                   
datalad publish --to=datalad-public -r --missing skip  12.32s user 5.71s system 44% cpu 40.903 total

$> datalad --version
datalad 0.12.5.dev606

yet to check where it gets ideas about uuid b38aead5-b9da-4fac-8588-286990fd2c94, but the point is that sourcedata isn't published to http://datasets.datalad.org/?dir=/labs/poldrack/NP-180740 . there is only an empty directory (as for any non "fulfilled" git submodule). So some check fails, it tries to copy data there for a while only to fail with various cryptic messages as per above.

[yarikoptic]

the situation is less exciting, but still worth accounting for: I had that subdataset published before (so it got its datalad-public remote), but then I had uninstalled it on the published remote side. But somehow locally it still has remote git-annex branch:

(git)smaug:…btrfs/datasets/datalad/crawl/labs/poldrack/NP-180740/sourcedata[master]git
$> git branch -a
  git-annex
* master
  remotes/datalad-public/git-annex
  remotes/origin/HEAD -> origin/master
  remotes/origin/git-annex
  remotes/origin/master
  remotes/origin/synced/master

$> git config remote.datalad-public.url
ssh://datasets.datalad.org/srv/datasets.datalad.org/www/labs/poldrack/NP-180740/sourcedata

$> ssh datasets.datalad.org ls -la /srv/datasets.datalad.org/www/labs/poldrack/NP-180740/sourcedata
total 0
drwx--S--- 1 yoh datalad   0 Mar 13 10:15 .
drwxr-sr-x 1 yoh datalad 312 Mar 13 10:15 ..

and if we just publish without specifying any paths, we do not even try to fetch it (to fail) but right away invoke git annex copy with bunch of files which screams in the back all those

| git-annex-shell: expected repository UUID d326b73e-1f9f-4ed0-97cb-1bbfd24f5057 but found UUID b38aead5-b9da-4fac-8588-286990fd2c94

where b38aead5-b9da-4fac-8588-286990fd2c94 is the annex UUID of the parent dataset on the remote side.

[yarikoptic]

Most likely datalad push would exhibit the same problem, didn't check, just assuming