- Use the actual BOM as first parameter instead of the whole harness.
- Use a whole AdditionalComponent as second parameter instead of each
  attribute separately.

Move the lambda declaration out of the function scope for common
access from two different functions.

Assign input designators once to a temporary variable for easy reusage.

https://stackoverflow.com/questions/4978567/should-a-return-statement-have-parentheses

- Use one common entry loop to consume iterator only once.
- Use same key function for sort() and groupby(),
  except replace None with empty string when sorting.

Seems to be the most popular search alternative:
 https://stackoverflow.com/questions/8653516/python-list-of-dictionaries-search

Raising StopIteration if not found is better than returning None
to detect such an internal error more easily.

Make a list from the group iterator for reusage in sum expressions
and to pick first group entry. The expected group sizes are very small,
so performance loss by creating a temporary list should be neglectable.

Alternativly, itertools.tee(group, 3) could be called to triplicate
the iterator, but it was not chosen for readability reasons.

This type alias describes the possible types of keys and values in
the dict representing a BOM entry.

Build output string in component_table_entry() as the similar strings
in generate_bom(). Repeating a couple of minor if-expressions is small
cost to obtain a more compact and readable main expression.

This way, both BOM and harness.additional_bom_items uses the same
set of keys in their dict entries. This was originally suggested
in a #115 review, but had too many issues to be implemented then.

to simplyfy code

as proposed in #222

zzz