Support nimble2 (with conflicts resolved)

[perrydv]

@paciorek Here is a PR with all the work to support nimble2 and with merge conflicts resolved.

[perrydv]

@paciorek This turns out to be a bit tricky. I fixed some problems like changing my code from using type to instrType in instrClass objects to catch up to your changes. Now I have test failures that don't obviously have to do with my contributions but I suspect come from some branch/merge glitching. Could you look at line 1616 of test-nimbleModel and line 455 of test-setupMargNodes?

[paciorek]

Ok, I am working my way through the test failures. I'll report on them individually here.

Initial side note: When I build the package from support-nimble2-resolve-conflicts, I see this

Warning in utils::tar(filepath, pkgname, compression = compression, compression_level = 9L,  :
  storing paths of more than 100 bytes is not portable:
  ‘nimbleModel/inst/include/nimbleModel/predef/modelValuesBase_nClass/modelValuesBase_nClass_copyFiles.txt’

[paciorek]

The test failure in test-setupMargNodes is from line 455, in particular the 3rd declaration, the declaration of y. It's happening because there is a (new, compared to master branch) element of varInfo that has name "NA". That element seems to appear because in make_modelClass_from_nimbleModel, in decl_methods |> lapply(\(x) all.vars(body(x))) , "i" is appearing as a variable related to getParam_one, and of course i is not in any varInfo:

$getParam_one
[1] "paramID_"                                            
[2] "LocalAns_"                                           
[3] "model"                                               
[4] "y"                                                   
[5] "idx"                                                 
[6] "lifted_b1_times_x_oBi_cB_plus_mu_oBi_cB_L2"          
[7] "lifted_exp_oPb2_times_x_oBi_cB_plus_mu2_oBi_cB_cP_L2"
[8] "i"                                                   

So I think this one does go back to changes you've made.

[EDIT] It looks like the test-nimbleModel.R line 1616 issue is the same thing. i has crept into declVars in the processing of the second declaration in make_modelClass_from_nimbleModel:

Browse[1]> declVars
[1] "y"                                                                          
[2] "n1"                                                                         
[3] "n2"                                                                         
[4] "mu"                                                                         
[5] "lifted_chol_oPpr_oBn1_oBi_cBton2_oBi_cB_comma_n1_oBi_cBton2_oBi_cB_cB_cP_L2"
[6] "logProb_y"                                                                  
[7] "pr"                                                                         
[8] "i"                                                                          

[perrydv]

Thanks for the help @paciorek. After dealing with a bunch of issues, this should pass tests on this run or soon. The last issue was just changing expect_identical to expect_equal in some spots, for ubuntu failures that pass on MacOS. So I'd say it's ready for any code review you want to do.

[paciorek]

Ok, I am merging in. At some point, I'd like to better understand the setup for getParam, but I'm not looking deeply into it now.