Optional location-related fields should not be written when empty

[arthurvd]

In the .ext file, locations for Lateral and Boundary objects can be specified in multiple ways, so all these fields are optional.
When an MDU+ext file is saved to disk, also all optional fields are written, which is not an error, but it is unnecessarily confusing, nonetheless.
Example .ext input:

[Lateral]
	id = lat_871
	name = lat_871
	type = discharge
	locationType = 1d
	branchId = riv_RS1_909
	chainage = 99.678
	discharge = boundaries.bc

[Boundary]
	quantity = waterlevelbnd
	nodeId = 141133.805000_395446.430000
	forcingFile = boundaries.bc

Currently saved output:

[Lateral]
id             = lat_871
name           = lat_871
locationType   = 1d
nodeId         =  # <<< May be omitted
branchId       = riv_RS1_909
chainage       = 99.678
numCoordinates =  # <<< May be omitted
xCoordinates   =  # <<< May be omitted
yCoordinates   =  # <<< May be omitted
discharge      = boundaries.bc
type           = discharge

[Boundary]
quantity     = waterlevelbnd
nodeid       = 141133.805000_395446.430000
locationfile =  # <<< May be omitted
forcingfile  = d:\dam_ar\dflowfm_models\HYDROLIB\MGB_FMonly\fm\saved\boundaries.bc
bndwidth1d   =  # <<< May be omitted
bndbldepth   =  # <<< May be omitted

Note that the same issue arises for Structures (locations) and DamBreaks (waterlevel locations)

[priscavdsluis]

@rhutten, do you know whether it is "safe" to skip empty properties for all files?
The MDU is the only INI file I can think of that is OK to write empty properties.
Please verify for which files this change can or cannot be implemented.

[priscavdsluis]

@arthurvd, I will change it to a bug, because we should not be serializing models to file that will raise error when deserialized again. Hope you agree with this :)

[priscavdsluis]

@arthurvd and @rhutten, I have made the assumption that empty properties should be written for the MDU file and they should be skipped for all other files. Please someone verify this.
Putting this issue for review.