trip dev 1.10.0

Remove tests expecting silence.
Fix situation where all date-times being NA resulted in 'Error: C stack usage 7975700 is too close to the limit'.

trip 1.9.0

filter_penSS() is removed, it didn't really work reliably and has no known users.
Updated DESCRIPTION, citation, help examples, removed disallowed warning suppression, added return value documentation, thanks to CRAN.
Updated to new spatstat.explore (removed spatstat.core).

Fixed documentation cross-references to sp topic, thanks to CRAN.
Migrate to reproj to avoid rgdal package. Ancient function tripTransform has been removed completely, and spTransform methods are now defunct and suggest use of reproj().
Removed problematic PROJ strings, thanks to Roger Bivand.
readDiag() and readArgos() now read all records from a single text vector after reading all raw lines. This means they can be used to read data from mixed text streams.
readDiag() no longer converts columns id and iq to factor, and gains arguments return_trip and read_alt to control output format.

BREAKING change: the sp package is now imported rather than full Depends. This means that calls to its functions like coordinates<-, CRS<-, proj4string<-, spTransform will need to be prefixed with sp::, or with the package explicitly attached with library(sp). This may cause some pain but should be offset by improvements in helpful ways to create trip objects.
The trip() function now accepts raw data frames or grouped data frames, with the first four columns assumed to be the x, y, time coordinates and the ID column. If the data frame is grouped the first grouping column is used as the ID.
When trip objects are created they are by default forced into basic data sense, and so will usually succeed on creation as long as the nominated coordinates, trip group ID and date-times are sensible. The data will be automatically cleaned up for 1) duplicated records 2) duplicated date-times 3) records out of order.
summary(trip) now does not calculate the rolling RMS speed for performance reasons, this might be reviewed in future.
New function write_track_kml to produce a compressed Keyhole Markup Language ('KML') file in 'KMZ' form, for use in Google Earth.

readArgos gains a new argument read_alt which is set to NULL by default and maintains existing behaviour. If set to 1 or 2 the corresponding "alternative" location is returned. 1 is a standardized location corresponding to the original PRV message, and 2 is a "dummy" location. Added by request of Chris Ward.
Now using geodist package rather than internal distance function. There's no speed benefit for moderately small data sets (1000s of points) but there is for much larger data. Some small numeric changes in distances when using longitude, latitude data - in trackDistance and homedist but at the level of metres in terms of precision.
New support to coerce to sf with as(x, "sf") which creates every segment as a LINESTRING. If sf::st_as_sf() is used standard coercion from sp -> sf is used and results in POINT.
Add trip() support including coercion from and to the amt package type track_xyt.
Add trip() support for trackeRdata objects from the trackeR package, mousetrap objects from the mousetrap package. Conversion to these types from trip is not practical.
Add trip()) support for sf objects with POINT geometry. The column names for date-time and id must be supplied. Z or M values from XYZ, and XYZM, and XYM are currently ignored.
Add partial trip() support for telemetry objects from the ctmm package, see #33.