Implements an iterative version of PLS based on the method by Milidi and Reteria,
"DPLS and PPLS: two PLS algorithms got large data sets." Computation Statisitics and Data Analysis vol.48 pp125-138 (2005)
Loads a ini style data file
general format
# a comment
[block name.sub block name:type]
variable1 = 56
variable2 = "a string"
variable3 = [13,23,12,14] # a vector
The resulting data is stored as a tree of node (starting with 'root')
Each node represents a block, containing named variables and child blocks
Each block has a 'type' associated with it (a string value) and a name.