error handling and submit_command refactoring (with tests for synthesis)

[ssteffl]

other noteworthy changes include:

• added lazytrycrossref and lazytrysubst meta types to hammer-config (with tests)
• hammer-ir keys for filtering to nldm and subset of stdcell libs for debugging
• added more sdc constraint hammer-ir keys, and to sdc generation
• added time_unit and capacitance_unit to sdc, and removed current workaround
• added is_physical key for synthesis, to toggle lef/def usage

other minor changes include:

• more readable syn.tcl formatting
• added CapacitanceValue unit for default_load
• added more timing/linting reports to genus flow

[jwright6323]

After an extremely quick glance this looks good, but I'll dive a little deeper tomorrow.

[edwardcwang]

Generally looks good overall, thanks for the PR!

One thing I might note is that there's a lot going on here - if possible, perhaps it could be possible to refactor this into multiple smaller PRs that are easier to look at in isolation?

[edwardcwang]

Suggested change

	if not(has_vt):
	if not has_vt:

[edwardcwang]

This is meant to be a suggestion for all 4 lines but I think GitHub doesn't know how to do it for all 4 lines...

Suggested change

	# how many output lines to buffer in memory. "" for unlimited
	max_output_lines: null
	# how many error lines to buffer in memory. "" for unlimited
	max_error_lines: null
	# How many output lines to buffer in memory.
	# If unset, buffer unlimited lines.
	# type: Optional[int]
	max_output_lines: null
	# How many error lines to buffer in memory.
	# If unset, buffer unlimited lines.
	# type: Optional[int]
	max_error_lines: null

[edwardcwang]

Suggested change

	self.assertEqual(db.get_setting("lazy.numbers2"), [1,2,3])
	self.assertEqual(db.get_setting("lazy.numbers2"), [1, 2, 3])

[edwardcwang]

How does this differ from crossref?

If "abc ${foo}" is trysubsted, how does it work if foo is an int or bool?

[ssteffl]

it breaks. i found no good alternative there.

but the common case is when a key refers to "${foo}", I want this to work regardless of what the type of the setting it.

the only difference between trysubst and subst, is in the subst_target() function:
the trysubst will simply return no targets if it detects a non str/List[str] setting value, which is what I want. I left the default behaviour of subst alone.

I would prefer to just have subst, and change the behavior to allow non-string setting values instead of raising an exception.

[edwardcwang]

Not sure as to why this is here...?

[ssteffl]

the HammerSubmitCommand will always drop a self-contained command script and a log file in the run-directory. In haste, i just hard-coded the command script to submit_command.sh. when i run tests, that file got dropped in the hammer work-tree.

if we ever want to run multiple hammerSubmitCOmmands in the same working-directory, this will have to be fixed (i'm assuming it will)