Changelog

pass through SlX formula in call
re-corrected #19 because the fitted model weights component may be NULL
suppress warning from multcomp::glht as the test which throws the warning is discarded

move expm from Imports to Suggests #42
added zero.policy pass-through to spdep::mat2listw calls in predict.Sarlm and to spdep::sn2listw in sids_models.Rmd; set spdep requirement to 1.3-1
corrected #19 because the fitted model weights component is never NULL, but may have a single unique value

raised #39, no support for weights in SEM/SDEM/SLX #39
address #37; #38 remains (no formula Durbin support for prediction using any Sarlm object)
address #19 by not reporting AIC where case weights are used in spautolm or errorsarlm
address bug in predict() for new data, SDEM. Others in #37, #38 need work.
Further added checking for SLX/SDEM impacts and edge/corner cases; starting transition to use multcomp in place og gmodels

Attending to SLX/Durbin/non-W-style weights: #7, #36, #26, #35, #30 #24, #23, partly based on #13

make local copy of gmodels::estimable() for lm objects only, add authors to package contributors
-Wstrict-prototypes fixes

updating coercion for Matrix 1.4-2
updating dontrun/donttest for package split (previously unchecked, mostly in aple)

#18 standardize use of coef() methods for (some) fitted model summary objects
https://github.com/tidymodels/broom/issues/1003#issuecomment-798694400 changing spatialreg model output class names: spdep sarlm -> spatialreg Sarlm, spautolm -> Spautolm, stsls -> Stsls, gmsar -> Gmsar, lagmess -> Lagmess, SLX -> , SlX, MCMC_s*_g -> MCMC_s*_G, SFResult -> SfResult, ME_res -> Me_res, lagImpact -> LagImpact, WXImpact -> WXimpact
#16 merged coordination of impacts methods (Gianfranco Piras)
#14 merged correction to SDEM and SLX impacts when a lagged intercept is present (Tobias Rüttenauer).

#2 Split spatialreg from spdep; spdep functions still present there as deprecated

Version 1.3-2 (development)