Drizzlepac

======================

- Changed the definition of Megabyte used to describe the size of the buffer
for create median step (``combine_bufsize``). Previously a mixed
(base-2 and base-10) definition was used with 1MB = 1000x1024B = 1024000B.
Now 1MB is defined in base-2 (MiB) as 1MB = 1024x1024B = 1048576B.

- Redesigned the logic in ``createMedian`` step used to split large
``single_sci`` images into smaller chunks: new logic is more straightforward
and fixes errors in the old algorithm that resulted in crashes or
unnecessarily small chunk sizes that slowed down ``createMedian`` step.

- Due to the above mentioned redesign in the logic for splitting large images
into smaller chunks, now ``overlap`` can be set to 0 if so desired in the
``minmed`` combine type. Also, it is automatically ignored (set to 0) for all
non-``minmed`` combine types. This will result in additional speed-up in the
Create Median step.

- Both ``AstroDrizzle()`` and ``TweakReg()`` now can be called with
``configobj`` parameter set to ``'defaults'`` in order to indicate that
values for the not-explicitly-specified parameters should be set to
the default values for the task instead of being loaded from the
``~/.teal/astrodrizzle.cfg`` or ``~/.teal/tweakreg.cfg`` files that store
latest used settings.

- Updated documentation.

========================

- Restore recording of correct ``EXPTIME`` value in the headers of
single drizzled ("single_sci") images. See
https://github.com/spacetelescope/drizzlepac/issues/93 for more details.

- Fixed a bug in ``drizzlepac`` due to which user provided ``combine_lthresh`` or
``combine_hthresh`` in the ``CREATE MEDIAN IMAGE`` step were not converted
correctly to electrons (processing unit). This bug affected processing of
WFPC2, STIS, NICMOS, and WFC3 data. See
https://github.com/spacetelescope/drizzlepac/issues/94 for more details.

- Modified print format so that scales, skew and rotations are printed with
10 significant digits while shifts are printed with 4 digits after the
decimal point.

========================

- Fixed a bug in expanding reference catalog in ``TweakReg`` that would result
in the code crashing.
See https://github.com/spacetelescope/drizzlepac/pull/87 for more details.

- Fixed a bug due to which user catalog fluxes would be interpreted as
magnitudes when ``fluxunits`` was set to ``'cps'``.
See https://github.com/spacetelescope/drizzlepac/pull/88 for more details.

- Fixed a bug due to which user-supplied flux limits were ignored for
the reference catalog.
See https://github.com/spacetelescope/drizzlepac/pull/89 for more details.

==========================

- Fixed a bug in computing optimal order of expanding reference catalog that
resulted in code crashes.
See https://github.com/spacetelescope/drizzlepac/pull/86 for more details.

==========================

- Fixed ``astrodrizzle`` lowers the case of the path of output images issue.
See https://github.com/spacetelescope/drizzlepac/issues/79 for more
details.

- Fixed ``tweakreg`` ignores user-specified units of image catalogs (provided
through the ``refcat`` parameter) issue. See https://github.com/spacetelescope/drizzlepac/issues/81 for more details.

- Corrected a message printed by tweakreg about used WCS for alignment. Also
improved documentation for the ``refimage`` parameter.

=====================

- ``drizzlepac.adrizzle`` updated to work with numpy >=1.12 when they implemented
more strict array conversion rules for math. Any input which still has INT
format will be converted to a float before any operations are performed, explicitly
implementing what was an automatic operation prior to numpy 1.12.