CRAN release version 1.0.20

What's Changed

• Correct residual variances when replacing interaction variances by @Kss2k in #137
• Add handling of composites for LMS approach by @Kss2k in #138
• Better optimization of mean structure by @Kss2k in #139
• Add composite/formative construct support to modsem_pi by @Kss2k in #140
• Add MC-LMS-CAT and MC-QML-CAT estimators by @Kss2k in #141
• Cache H0 using memoise by @Kss2k in #142
• Penalized standard errors by @Kss2k in #143
• Refactor some LMS code by @Kss2k in #144
• Optimized gradient lms by @Kss2k in #146
• Better handling of composites in LMS by @Kss2k in #147
• Parallel E-step by @Kss2k in #148
• Improve modsem_predict.modsem_da() by @Kss2k in #149
• Use lavaan standard for condition messages by @Kss2k in #150
• Optimize gradient/hessian calculation for cov model in lms by @Kss2k in #152
• Add se argument to modsem_predict.modsem_da() by @Kss2k in #153
• Standardize residual covariances between endogenous lvs by @Kss2k in #154
• Optimized Gradient/Objective Calculation for QML by @Kss2k in #155
• Better formatting of messages by @Kss2k in #157
• Make ordered.se="delta" the default by @Kss2k in #158
• Fix bug in #156 by @Kss2k in #159
• Better message formatting by @Kss2k in #160
• Allow residual covariances between exogenous and endogenous latent variables in the LMS approach by @Kss2k in #161
• Handle residual covariances between endogenous and exogenous lvs in modsem_predict() by @Kss2k in #163

Full Changelog: v.1.0.19...v1.0.20