Fix compression components

We were mistakenly comparing the size of the working buffer to
the input size and not compressing if the working buffer was
larger. However, zlib always uses a larger working buffer,
and so this prevented us from actually compressing.

Compare the size of the resulting compressed data to the
input data and don't use the compressed block if it isn't
smaller - it can happen, though probably rare.

Signed-off-by: Ralph Castain rhc@pmix.org
(cherry picked from commit 0fe0c59)

Use the correct value for the number of info to unpack

Correct a typo

Signed-off-by: Ralph Castain rhc@pmix.org
(cherry picked from commit 58f49a1)

Remove stale/unused tests

Don't believe anyone ever ran the "test_v2" collection, and
no indication that they are currently being exercised.

Signed-off-by: Ralph Castain rhc@pmix.org
(cherry picked from commit a9ef77b)

bitmap num_set boundary condition bugfix

Signed-off-by: Matthew Whitlock mwhitlo@sandia.gov
(cherry picked from commit 08c22de)

preg/compress parsing bugfix

Signed-off-by: Matthew Whitlock mwhitlo@sandia.gov
(cherry picked from commit 2dfa69c)

Port bug fixes to zlibng component

Port the zlib component changes in PR #3708 to
the zlibng component as that is the default
in use for newer RH (and affiliate) images

Signed-off-by: Ralph Castain rhc@pmix.org
(cherry picked from commit bd68463)

Flush namespace sinks' residuals before destroying them

Signed-off-by: Matthew Whitlock mwhitlo@sandia.gov
(cherry picked from commit f773947)

Put the sink cleanup in the sink destructor

As per the discussion on the PR, move the sink cleanup
calls to the sink destructor.

Signed-off-by: Ralph Castain rhc@pmix.org
(cherry picked from commit f04adb2)

Replace sprintf with snprintf

Per note in the OMPI project, at least one compiler family
is removing the "sprintf" function. Replace all uses of
that function with the safer "snprintf" version.

Signed-off-by: Ralph Castain rhc@pmix.org
(cherry picked from commit 1c63aea)