Fix default `projargs` and deploy range-edge indices

Two changes:

• Fix the default value for projargs when plotting to avoid projection issues with global coastline shapefile
• Change default settings to now calculate and plot range-edges.

Adding `predict` and marginal effects plots

Added functionality includes:

• predict S3-class generic function enters beta testing
• Integrating effects package to facilitate plots of marginal covariate effects
• Re-adding continuous integration testing of core functionality

See NEWS.pdf for more details

hotfix for typos affecting backwards compatibility and little used features

Typos identified by Arnaud Gruss affecting:

• Backwards compatibility
• Different specification for vessel effects
• Different specification for spatially varying responses for the two linear predictors
  where any issues should have resulted in model-build errors, and therefore likely had little impact on users.

Improved covariate interface and spatially varying catchability

VAST release 3.6.0 includes new features:

• NEWS document
• Improved covariates interface
• Spatially varying catchability effects

Hotfix for previous minor release...

... fixing location of brackets (also having actually running the original example and integrated-tests)

Hotfix for crash...

... occurring when specifying density-covariates using formula interface, specifically when including negative values (for reasons that were not captured in current set of integrated tests). Thanks to @TiffanyVidal in issue #251 for showing the bug in a simple way, and also identifying the issue.

Hotfix for macOS compile issue...

... eliminating use of std::complex, which causes errors in some macOS compilers for reasons unknown. Thanks Dan Cushing for using his machine to test fix. This fix eliminates a rarely used feature for estimating interactions with bounds on species fluctuations (VamConfig[1]=3)

Hotfix for bug in issue #246

Fix bug in VAST::make_data resulting in obviously nonsensical results in cases where response b_i contains some NA values and when Options['treat_nonencounter_as_zero']==TRUE (i.e., an index-standardization model). See #246 and thanks @Cole-Monnahan-NOAA for flagging the problem.

Fix compile bug...

... as documented in #242, changing abs(.) to fabs(.) in CPP version >= 9.3.0

Barrier-SPDE and cohort effects

VAST release 3.5.0 involves :

• Adding feature for barrier-SPDE approach;
• Changing density covariates to index by X_gctp to allow treating cohort effects via the spatially-varying coefficient interface
• Eliminating the previous, deprecated interface to seasonal effects to simply CPP code