1.43
	
	- New calibration document and FITS file structure documents added
	to doc/ sub-directory inside the crush distribution. Later, these
	documents will be made available via the CRUSH website
	(http://www.submm.caltech.edu/~sharc/crush), once their contents
	are more or less finalized...

	- Updated README, with a new section added for deep field or very
	faint sources, reduced by the -deep option.

	- Internal smoothing of maps (via -smooth=internal:X option or
	MAP_SMOOTHING=internal:X key) now takes place only until
	faithful map construction is activated (-Ifidel= option or
	MAP_FAITHFUL_TURN key). This way the adverse effects of internal
	smoothing are kept minimal.

	- New option -scanMaps (WRITE_SCAN_MAPS) will write maps for all 
	individual scans as well as the composite. The scan maps will be 
	placed in the default location (REDUCED_MAP_PATH), and their file 
	names are the standard [source name].[scan #].fits format. These files 
	do NOT contain scan specific data. Look for those in the composite 
	image.

	- Slightly improved source extraction confidence-to-sigma calculation.

	- SourceMap typical lfux reporting changed for better pipeline
	ordering.	

	- Remeasured 350um tau A,B values based on 2003sep-2005may calibrator
	data (315 independent calibrations). The best fit values are:
 		A(350um) = 29.41 +- 0.56	
		B(350um) = -0.417 +- 0.06
	The conversion factor is measured to be 353+-13 nV/(Jy/detector) 
	for Uranus.
 
	- SkyPolynomials subtracted from unflagged pixels only. Now, it 
	correctly subtracts from all pixels.

1.43-a1 

	- MAP_SMOOTHING key and -smooth= flag behaviour changed. The new syntax 
	is 
		[directive:]value
	
	where the directive can be 'internal', 'final' or 'always'. Omitting
	the directive assumes 'always'. This way two different smoothing values
	may be defined. Use 'internal' for the smoothing withing the reduction
	cycle. This allows decreasing the fitted degrees of freedom, and help
	get rid of misbehaving pixel trails. However, can cause residual map 
	artefacts for bright sources due to the 'imperfect' source removal 
	due to the effective beam widening from the smoothing.
	The 'final' directive sets the smoothing on the final maps, at the
	end of the reduction. Configuration files have been adjusted.

	- Pipeline reordering (-reorder/-noreorder) automatically invokes
	the '-bright' option (i.e., parses bright.cfg) when the source
	precedes the correlated noise in the reordered pipeline. This results
	in much better default reduction of bright sources (e.g. Jupiter, Mars)

	- New minimalistic bright.cfg (invoked by -bright) for 1000+ Jy
	sources.

	- MAP_ROBUST_WEIGHT key and -mapWeight= flag introduced to specify
	via a true/false argument whether to estimate robust weights of maps
	based on the observed vs the expected scatter. While it works well
	for most cases, one might one to disable it for the case of very
	bright sources (with strong beam lobes) or brightish but extended
	sources.

	- (!) Scan weights derived from map scatter were erroneously applied
	twice.

	- [default pixel weights were off by a large factor...]

	- ROTATOR_ZERO (-rot0=) 'auto' will now correctly identify Jan 2003
	scans with 60.6, as well as experimental Nasmyth configurations 
	28-30 Aug. 2004. Also, SHARC2_PIXEL_SCALE (-pixelScale=) 'auto' will
	set the proper plate scales for those dates, while 
	CHOPPED_OBSERVATION='auto' will reduce correctly Nov. 2002 data 
	without the chopper signal (which was not present at the time). 
	This is all with the intention of providing better defaults.

	- Revised and cleaned up gain calculation algorithms. Recalibrated
	nonlinearity coefficients, separately for the original configuration, 
	and for after the filter replacement. New configuration key 
	NONLINEARITY_COEFFICIENTS and option flag -pixelCoeffs= can be used
	to specify the nonlinearity file to use. The setting 'auto' should
	always chose the correct file automatically.
	Using the correct pixel coefficients also makes direct tau measurements
	more usable via TAU_SOURCE = direct (or -tau=direct).

	- The previous version shipped with a non-standard 'calibrate' utility.
	Now the utility has been restored to its original form.

	- Updated 450um & 850um tau definitions with Archibald's and Darren's 
	values.

	- Map filtering + 3rd order sky polynomial combo substituted with
	RegionalCorrelations + linear order sky polynomial. The two yield
	nearly identical results. The first one is faster but the second option
	is better characterised for its filtering effects (e.g. taken into
	account by 'detect'). Thus, it is a matter of preference.

	- New reduction option -noflag introduced to disable all flagging.
	This is effectively just a convenient shorthand for 
	'-Ispike=off -Itemporal=off -Ispectral=off'.

	- Flux scale was widened to accomodate larger label width.

	- Simplified PixelDrifts timeconstant syntax to 'auto' or integer, 
	eliminating the possibility of individual bolometer timeconstants.
	The individual time constants, while a noble feature, were not
	particularly useful due to the difficulty of measuring each detector's
	time constant in a reliable way. As a result the relevant code is
	a lot cleaner and shorter...

	- Revisited and corrected PixelNonlinearities model. Now works 
	semi properly, although the derived nonlinearity parameters don't 
	stick. When active, it is also solved for with initial gains.
	Use with caution if at all...

	- Changed pixel flagging automatic activation chi^2 requirement to be
	less than 2.0. This affects ResidualSpikeRemoval, SpectralPixelTest,
	TemporalPixelTest.

	- Residual opacity adjustment now happens at source extraction
	rather than after re-derivation of the correlated noise.

	- Adjusted 13.4Hz filter width to 0.3 Hz.

	- [Explicitly resolved row-polynomial and pixel-drift degeneracies
	with correlated noise, s.t. residual adjustments to correlated
	noise cannot pollute these models. Should result in better pipeline
	ordering for faint sources...]

	- [frame.offset was not correctly synched with correlated. Fixed. ]

	- New tool 'detect' added to aid source extraction from SHARC-2
	maps. The tool assumes that the underlying map noise is Gaussian. 
	To ckeck that this is true, use 'histogram', or other means... 

