Asammdf

Fixes
* fix crashes triggered by selecting the common axis when channels have been removed from the selection list

Fixes
* fixes 248
* fix error when dragging channel groups that contain no channels
* show correct information about the found CAN ID's during extraction
* remove invalid chars from output csv file name
* missing return value in blocks.mdf_v4._load_signal_data (thanks to JulienGrv )


Changes
* _load_data containing invalidation bytes
* improve C extension speed for unsorted mdf v4 files

Fixes
* improve drag & drop support
* fixes 243, 244, 245
* fix filtering of byte arrays in the tabular window
* fix fast parsing of structure channels


Changes
* C extension optimizations for parsing VLSD channels and sorting files
* add option to use just the field names in get_group and to_dataframe

Changes
* implements 238
* performance improvements
* add **copy** argument to _Signal.validate_ method
* use _select_ in _to_dataframe_ (lower memory usage)
* improve drag and drop support in the GUI

Standalone GUI builds
* [Linux](https://ci.appveyor.com/project/danielhrisca/asammdf/builds/27926742/job/ku0aul2xs4omgqo9/artifacts)
* [Windows](https://ci.appveyor.com/project/danielhrisca/asammdf/builds/27926742/job/td0jtbam9hwdg34h/artifacts)

Appveyor fix release

Fixes
* improve robustness to out of file addresses
* fix parse error issue for header comment
* don't change the current working directory when extracting attachments
* don't crash if the CAN database cannot be parsed by canmatrix
* keep original timestamps in the tabular view
* fix bus logging measurement when the channel group has no source
* missing channel_group.acq_name after appending signals
* remove unwanted name attribute in v4_blocks.ChannelGroup
* fixes 223, 224, 228, 233, 234
* fix handling of invalidation bits in select method
* ignore errors in text decoding
* handle duplicated field name sin structure channel composition
* fix handling of 0 cycles channel groups
* fix using display names
* fix in get data blocks info

Changes
* ignore parsing errors of channel conversions
* add measurement header comment to info section
* add possibility to remove channel name prefix in the tabular view
* add argument for to_dataframe to not perform interpolation and fill with NaN's
* implements issue 231
* rename 'time' to 'timestamps' in pandfas data frame and csv exports; makes it compatible with AWS