[FIRRTL][Dedup] Merge location information (llvm#2634)

This adds support for fusing location information when two modules are
deduplicated with each other.

[FIRRTL] Add test of numeric class, NFC

This is an additional test of a bug fixed in fe2fdb7 where an
annotation with a numeric class (e.g., "class":"42") could cause a crash
due to a blind cast to a StringAttr.  This locks in the behavior that
we ignore a numeric class instead of crashing.

Signed-off-by: Schuyler Eldridge <[email protected]>

[FIRRTL] handle testbench dir annotation in combination with dut anno…

…tation (llvm#2382)

Get files in the correct directory.

[FIRRTL/ExpandWhens] Support aggregate type registers (llvm#2305)

This commit changes ExpandWhens to handle aggregate type registers.
We need to expand the connection into individual ground type elements.

[TableGen] Factor common defs out into common `Types.td` (llvm#2120)

Create a CIRCT-wide `include/circt/Types.td` file that can collect the
common type, attribute, and constraint definitions used throughout the
CIRCT project, which may not be specific to one of the given dialects.
Things that land here would generally be candidates for upstreaming to
MLIR, but it's better to have a common place in CIRCT first where they
can incubate.

While we're at it, get rid of some of the type definitions which have
since appeared in upstream MLIR, and factor some first commonalities out
into `Types.td`. There will be quite a few more in the future.

[PyCDE] Remove export_verilog call from System.run_passes.

This was previously needed to get the LegalizeNames side-effects
before applying placements. Now that we apply placements and use
instance name symbols before LegalizeNames and ExportVerilog, it is
unnecessary.

[Python] Add bindings for hw::ParamVerbatimAttr. (llvm#2051)

This just needs to be plumbed up to Python.

[FIRRTL] Add filename option for seq mem metadata emission (llvm#2005)

Support setting the filename for the seqmem metadata.

[Python] Flush error stream in MSFT tests. (llvm#1981)

I found that this test consistently failed because the === Errors:
label was always showing up at the end of the error file being
checked. By manually flushing, the test consistently passes.

An issue is opened for a long term solution.

[HandshakeToFIRRTL] Add support for index_cast, zexti and trunci ops (l…

…lvm#1865)

zexti and trunci operations generate handshake components which wrap around FIRRTL `pad` and `bits` operations. The builder for each of these ops is used to support `index_cast`, which pads or truncates an input type compared to the fixed index-width (64 bits).