avoiding unecessary formatting.

This is done to prevent confusion when using the server/client interface.

format to allow a better merging of data with client.merge_data query
issue is that a MplData object is created on the server side silently...
so the user need to be warned.

client-sice >: fig.update(rawdata=['some_data'])

now should work.

declare aliases or addAxes (which make less sence but what ever...)

tested (by hand -- no unittest and no tutorial) and pass

Tests and tutorials are passing but clearly need some cleaning...

details and improve coding.

remark I should already inverse the way axes are appended (from bottom to top)
this is really not natural we are used to top to bottom (fucking matplotlib idiot!!)

So need to rethink the layout function.. anyway not a good name.

merged in issue 44.

Now the keyword aliases is treated differently. MyFig.update does not
replace the aliases but updates it. To allow the user to gives arguments
without caring of keep track of the former aliases.

to overwrite the attribute aliases completely the user now need to do:

fig.aliases = replacement

instead of

fig.update(aliases=replacement)

But not yet MyAxes and MyFig because need the server version

suppress the reinitialisation of fig.aliases in declareAliases

If the Axes is of type MyAxes but deriving from Another class It won't pass the
isinstance(ax, MyAxes) test

However it will pass
issubclass(ax.__class__, MyAxes)

But if ax is derived from MyAxes from another namespace:
ax.__class__ = some_namespace.MyAxes
MyAxes = some_other_namespace.MyAxes
It won't pass neither...

Therefore I commented this test before I find a solution

size of the figure.

now can use in mytool module

	usage()
or
	usage(<fct-name'>)

(this time with changes)

MyAxes.update() allows to write the MyFig.update() shorter and clearer. I vote to make use of that.

- Now the figure calls the update function of axes
  This is cleaner and object oriented.

def update(self, **kwargs):

    for ax in self.get_axes():
        ax.update(**kwargs)

BUG: two axes figure (test.figure3) shows only one axes

- all tests go through
- Now if data does not exists set to None