run_fixed inferred a fixed length shift register from a chain of FDRE
flops but assigned it ENPOL 2, which the cells_map template reads as no
enable and ties the enable high. FDRE has an active high clock enable, so
the shift register shifted every cycle and ignored stalls. Give FDRE and
FDRE_1 chains ENPOL 1 so the active high enable is kept.
Add tests/arch/xilinx/xilinx_srl_enable.ys covering the FDRE path.
Move the operand commute that puts the wider operand on A ahead of the
min-width checks. Checking the min-widths first made an asymmetric rule
(DSP_A_MINWIDTH != DSP_B_MINWIDTH) match only products that already had the
wider operand on A; a product in the other order was rejected before the
normalization.
Also `min` and `max` ops. RISC-V uses IEEE 754-2019 semantics where `min(+0,-0) == -0` and `max(+0,-0) == +0` so we do the same here. We could make it optional, but as I understand it the newer behavior is still backwards compatible (since previously it was valid to have selected either).
Switch inputs back to `anyseq`, and add coverage for muladd with constant multiplier output and varying addend (also an assertion).
Use an `ifdef` to control clocked properties (because of the assertion, it's no longer just covers that are clocked).
Coverage supports `sqrt`, including new general rounding detection instead of just inf/ebmin/zero (since they aren't possible with `sqrt`).
More `sqrt` assertions, as well as the addition of `altsqrt` verification.
Some adjustments of macros.
Include mask/map for abc inputs (and switch to `anyconst` instead of `anyseq`).
Add false divide check for mantissa.
Covers aren't currently being tested by anything (and have to be removed for `sat`), but I've been using it locally with SBY to confirm that the different edge cases are able to be verified (e.g. when verifying HardFloat against symfpu while using the masked inputs to reduce solver time).
