1.2.0

• Introduce the new parameter output.append_mode to select behaviour
  when appending new data which overlaps in the append dimension with the
  existing data.

• Adjusted the built-in pre-processing to handle more cases:
  Bounds dimensions of concatenation coordinates are extended by their
  dimension and transformed to variables if necessary.

• Whether to adjust output metadata (global attributes) after the last
  write/append can now be forced by the new output.adjust_metadata
  setting whose default is false. If set to true, this will adjust
  the following metadata attributes:

  • "history"
  • "source"
  • "time_coverage_start"
  • "time_coverage_end"

  In addition, extra metadata (global attributes) can now be added
  after the last write/append using the new setting
  output.metadata whose value is a mapping from attribute
  names to values.

  The above functionality is also reflected in two new CLI options
  --adjust-metadata and --finalize-only. In combination, they
  can be used to later adjust metadata in already existing Zarr
  datasets. (#20, #34)

• Fixed a problem that avoided appending datasets that contained variables
  with dtype = "|S1" (ESA SST CCI). Such variables are written initially,
  but then no longer appended at all given that they do not contain the
  dimension along we append. Otherwise nc2zarr will still fail. (#38)

• Fixed a severe problem where data was "encoded" twice before
  being appended to existing data. The bug became apparent
  with input.decode_cf = false (the default) and for variables
  that defined a scale_factor and/or add_offset encoding
  attribute. (#35)

• Added a CLI tool nc2zarr-batch.

• Add and update some sample configurations and scripts.