Added support for `K` state elimination

[dlcole3]

Added support for both sparse and condensed formulations of the K state elimination matrix.

• Added the K matrix as an attirbute of the LQDynamicData
• Added function for building the J1 matrix in the sparse format when K is defined so that it includes the K constraints
• Added two new functions for building the condensed blocks and the condensed G block for the condensed formulation since the algebra is signficantly more complex with K defined.
• Separated _build_condensed_lq_dynamic_model into two functions, one for when K is defined and one when it is not. These functions are called separately in LQDynamicModel.
• Added new tests for the K matrix formulation. Added tests for when K is defined with bounds, when K and S are defined with bounds, when K is defined without bounds, and when K and S are defined and not all u variables are bounded.
• Updated the sparse_lq_test.jl file so that it forms the JuMP.jl model with the K matrix.

[dlcole3]

Code needs a lot of revision. Need to use multiple dispatch. I am closing this PR and will open a new one when I have revised the code.