Previously, if you uploaded a file as an SLO then re-uploaded it
with the same segment size and mtime, the second upload would
go delete the segments it just (re)uploaded. This was due to
us tracking old_slo_manifest_paths and new_slo_manifest_paths
in different formats; one would have a leading slash while the
other would not.

Now, normalize to the stripped-slash version so we stop deleting
segments we just uploaded.

Change-Id: Ibcbed3df4febe81cdf13855656e2daaca8d521b4
(cherry picked from commit 9021a58c240e156f54ffafdc4609868f348d3ebc)

This is a combination of three commits from master:

---

Isolate docs requirements

...since modern sphinx won't install on py27.

While we're at it, clean up some warnings and treat warnings as errors.

Also, fix up how we parse test configs so we can run func tests.

Related-Change: Id3c2ed87230c5918c18e2c01d086df8157f036b1
(cherry picked from commit 113eacf3b80f61d366b3e95b558b40f82ff728a4)

---

Drag forward prettytable in lower-constraints

Apparently version 0.7 got unpublished recently.

(cherry picked from commit 5bd66947fc3d8987d4b24d5119a346031004229e)

---

docs: Clean up formatting

(cherry-picked from commit 7563d9cb5642c85a31708fd863e2737f9d6c9419)

---

Change-Id: I3c6bca460ee5ff99195495a29e8b8bbf4ed804ff

1. update hacking version to latest
2. fix pep8 failed

Change-Id: Ifc3bfeff4038c93d8c8cf2c9d7814c3003e73504

Exposes the delimiter parameter, which the Swift API supports for
container listings.

Change-Id: Id8dfce01a9b64de9d1222aab9a4a682ce9e0f2b7

While investigating the failures when you move func tests to py3, I
noticed a whole bunch of

   ResourceWarning: unclosed <socket.socket ...>

noise. This should fix it.

While we're at it, make get_capabilities less stupid.

Change-Id: I3913e9334090b04a78143e0b70f621aad30fc642
Related-Change: I86d24104033b490a35178fc504d88c1e4a566628

There were two basic problems:

  - We'd try to import on every attempt at getting auth, even when we
    already know keystoneclient isn't available.
  - Sometimes devs would hit some crazy import race involving (some
    combination of?) greenthreads and OS threads.

So let's just try the imports *once*, at import time, and have None
sentinels if it fails. Try both versions separately to decouple
failures; this should let us support a wider range of keystoneclient
versions.

Change-Id: I2367310aac74f1b7c5ea0cb1a822a491e4ba8e68

Fix unicode handling in Python 3 and Python 2. There are currently two
failure modes. In python 2, swiftclient fails to log in debug mode if
the account name has a non-ASCII character. This is because the account
name will appear in the storage URL, which we attempt to pass to the
logger as a byte string (whereas it should be a unicode string). This
patch changes the behavior to convert the path strings into unicode by
calling the parse_header_string() function.

The second failure mode is with Python 3, where http_lib returns headers
that are latin-1 encoded, but swiftclient expects UTF-8. The patch
automatically converts headers from latin-1 (iso-8859-1) to UTF-8, so
that we can properly handle non-ASCII headers in responses.

Change-Id: Ifa7f3d5af71bde8127129f1f8603772d80d063c1

I'm giving up on trying to back out all of the test-requirements
up-revs, but let's try to stay compatibile with old requests/six.

As part of that, only disable some requests warnings on new-enough requests.

Note that we should now be compatible with distro packages back to
Ubuntu 16.04 and CentOS 6. Our six is still too new for Trusty, but
hey, there's less than a year left on that anyway, right?

Change-Id: Iccb23638393616f9ec3da660dd5e39ea4ea94220
Related-Change: I2a8f465c8b08370517cbec857933b08fca94ca38