[Edserplo_users] Update covers a number of bugs

[Steve Loch]

Bugs
The following bugs have been fixed.

B#503: Unable to read file conforming to ASCII General specification
B#175: Axis number not constrained  to be positive integer in the range 1-9.
B#290: Segmenting casts not done properly  (also affects Serplo for which the correction has also been applied) 
B#485: Initial selection with 'v' corrupted by series selection
B#180: Series without time channels allowed on TimeSeries page
 
For the last item the icon is now greyed out if there are no series with time channels. Typically these series are CTDs.  If there is a mixture of series with and without time channels there is no longer an abort when plotting on the associated page. 

The following is included for those who need to know where the time goes on a "nearly finished" product. Note that the vertical axis has changed.

Scale and Offset
The scale and offset have been handled throughout Edserplo, believe it or not, as Strings so that before they could be used for anything they had to be converted to numeric form and this conversion occurred in a very large number of places (like 20 or 30!).This was exceptionally clumsy and added considerable code clutter.  I've done most of the work to rectify this. Alterations in the String value are now converted to binary by a set of focusLost methods. Very similar considerations also apply to colour and to selection in the ChanPlo class. These are items  which I have yet to deal with fully. 

Vertical Axis
I was also forced to attack the problem of the vertical axis - because it was difficult to debug the code when implementing the changes outlined above.  The vertical axis algorithm (i.e. deciding the spacing of tick and annotation marks)  has failed on occasion. Unfortunately Ravi  didn't understand the Serplo code and, without asking,  implemented his own (rather bizarre) algorithm. It also used the LFM for formatting - which is over precise - instead of deducing the appropriate format for annotation directly from the range. I've now implemented the original algorithm, virtually unmodified - in about a day - and posted it in the super class so it is no longer duplicated between classes. I've since modified it so that annotation marks are more prominent than the tick marks. 

The annotation is not yet as it should be; bear in mind I had the Fortran formatter in Serplo.  I'm waiting to convert to Java 5.0 (=1.5) to get my hands on a proper formatting class which I think will save me the effort of coding my own. 

Horizontal Axis
This has been moved to the super class thereby removing 300 lines of duplicated code.

Upcoming changes
This leads on to the main aim which is to improve:- 

(1) processing of settings files - which is still wrong for tidal work (see B# 420,481) - and which involves complicated interactions with various other elements of the program and
 
(2) the storage of channel information, some of which is serialised to forms which are deprecated for long term storage. In other words as you move to a different version of Java you lose the ability to read your dump and settings files.

The major part of this work will commence after this update. This is a necessary step to phasing out Edteva altogether. 

Settings files will then need to be redone and dump files (not Edteva dump files which in this context are simply examples of another format) will no longer be readable.  

Steve