improved module code checks

[CeresBarros]

• Code checks that can have pervasive and serious consequences for caching, module order, etc., (like the second one highlighted in module code checks report possible issues Biomass_regeneration#17) should be made warnings;
• improve detection of object changes/assignment (see below)

[achubaty]

some messages (which should be warnings) are being triggered spuriously:

BC_HRV_preamble: inputObjects: NAME_1, LU_NAME are used from sim inside InitStudyArea, but are not declared in metadata inputObjects

Here I'm using e.g., sim$canProvs$NAME_1 (where canProvs is an sf object) but it's being flagged incorrectly.

Likewise, per PredictiveEcology/LandMine#4 (comment):

LandMine: inputObjects: ros is used from sim inside fireROS, but is not declared in metadata inputObjects

safe to ignore: ros is column in a data.table and sim$ROSTable$ros <- foo is used, so code checks are incorrectly flagging this.

[achubaty]

from PredictiveEcology/LandMine#4 (comment)

fireReturnInterval, sppEquiv are declared in metadata outputObjects, but are not assigned in the module

• fireReturnInterval is updated via sim$fireReturnInterval[] <- foo, so can ignore this message
• sppEquiv is a data.table and is updated by reference sim$sppEquiv[, LandMine := foo]

@eliotmcintire i'm not sure how easy/quick it is to update the code cheks in SpaDES.core to account for these. obv low priority, but now this is documented :)

[achubaty]

from fRI-Research/LandWeb_preamble#3 (comment):

LandWeb_preamble: outputObjects: TSFLayerName are assigned to sim inside InitMaps, but are not declared in metadata outputObjects

spurious: TSFLayerName being used as index to sim to reference another object:

ml[[TSFLayerName]][] <- as.integer(ml[[TSFLayerName]][])