ICAL Testing (instruframecal vsn 6.7)
-------------------------------------

Below we present a summary of ICAL testing and upgrades. We mainly validate:

- new calibrations/params in:
  /wise-ops/01/wise/fmasci/ical_bckup/ifr_ops/012010/
- mitigation of "global" quadrant-to-quadrant droop,
  *not* the intra-quadrant split phenomenon.
- other bug fixes, see below.

Summary of bug fixes to instruframecal after cover eject
(crucial to ensure all else works right): 

DROOP
-----
The correction here is relative to long-term reference pixel medians
for each quadrant ($refbase* parameters). If the fraction of good pixels is
<= $gfrac, the corrections are set to 0. Values will be biased since
we expect the real reference median to be < 0 => correction will be
underestimated if applied. Also we set a flag to fix these in the 
"drooprefine" routine at the end. The final droop refinement only occurrs if
the drpflag was set and there were quadrants with an insufficient fraction of
good ref pixels (<= $gfrac). We compute (any) relative residual offset
using the active pixels in neighboring quadrants at right/left or bot/top
(center array strip) whichever has a ref region with good fraction of
pixels (i.e., has $gflagQ=1), otherwise we give up. We don't use pixels in 
horizontal strips since bad split cases could bias results.

For example outputs, see jpgs under:
/wise-ops/01/wise/fmasci/FlightAnalyses/ical_test8/
W3 L0:  W3lev0.jpg 
W3 L1a: W3lev1a.jpg
W4 L0:  W4lev0.jpg
W4 L1a: W4lev1a.jpg

fmasci@caustic;rhe4(dev)% svn commit -m "related to vsn 6.7: testing of droop
algorithm revealed code had to be rearranged, optional droop refinement is now
performed at the end. FOR v3.5 OR EARLIER"
Sending        appcal/instruframecal
Transmitting file data .
Committed revision 6982.
fmasci@caustic;rhe4(dev)% svn commit -m "vsn 6.6: Implemented prelim
quad-to-quad droop correction with new params: gfrac and refbase* read from
wise-meta.tbl. Also disabled call to old (ground) droop algorithm. FOR OPS
v3.5 OR EARLIER. Tested in dev under:
/wise-ops/01/wise/fmasci/FlightAnalyses/ical_test6"
Sending        appcal/README
Sending        appcal/instruframecal
Transmitting file data ..
Committed revision 6969.

100% NAN FIX for QA
-------------------
vsn 6.5: For v3.5 or ealier: avoid choking when whole image is NaN'd, e.g.,
w3 and/or w4 due to anneals, the plan is still to have all saturated pixels
NaN'd at the end. Dummy histograms (with dummy in their title) will be made
if this occurs. Also, fixed a bitwise precedence warning from PDL

WRITE CALDIR TO HEADERS
-----------------------
vsn 6.4: FOR V3.5: propagate input cal dir used to output FITS headers

NON-LINEARITY BUG+RELATED
-------------------------
fmasci@caustic% svn commit -m "vsn 6.3: FOR 3.5 OR EARLIER: nonlinearity
calibration had a bug that caused large DN values (close to dynamic range) to
be turned into NaNs because discriminant in quadratic solution came out
negative. These rouge NaNs wrecked havoc in glitch detector downstream,
whereby -NaNs (yes, with a negative flavor) were causing bit 31 (the sign bit)
to be erroneously turned on in the frame masks. Why? It's the first time we've
seen real data and subject it to the non-linearity correction model";
Sending        appcal/instruframecal
Transmitting file data .
Committed revision 6816.

F. Masci, 01/21/2010