Parameterized loss function in curve fitting with new function .fitCurve2

[ChristopherEeles]

• Refactoring .fitCurve to .fitCurve2 with parameterized loss function
• Add optim method to .fitCurve2
• buildComboProfiles: Allow to build a drug combo profile without combo_viability in tre
• Further refactoring .fitCurve2
• Add helper methods to fix function parameters and collect parameters into a single list
• Add more optim helper methods to reformat general R functions into functions optimizable via stats::optim
• Add first attempt at .meshEval2 function
• .fitCurve2 done, testing vs .fitCurve
• Fix bugs in .normal_loss and .cauchy_loss; added unit tests for new loss being equal to .residual
• Loss function equivalence tests passing
• Added tests for .fitCurve and .fitCurve2; tests passing
• Add usage example to .fitCurve2
• Modified fitCurve tests to run within a reasonable time frame
• Add optim_only option to .fitCurve2
• Pre-merge with master
• Remove make_optim_loss helper
• Regenerate NAMESPACE and man