PSY parsing changes post package split

[luke-kiernan]

I had Claude audit the commit history of PSY's main for parsing changes that didn't make their way into PFFP.

There were a few things that clearly looked like bugfixes, so I ported them over: #51, #52, and #53 .

Probably should be ported over:

• 5b298a811 / f5e9f0dd4: PFFP's ramp_limits ignores base_conversion. Its docstring calls this out as a deliberate verbatim port of PSCB's inconsistency. PSY has since ruled it a bug and threads base_conversion through all four branches. Every other limit in _gen_base_and_limits is scaled, so ramp limits are the lone unscaled quantity whenever mbase != sys_mbase.
• b3542ea98 / 297a4759e: PSY moved VSC control modes to VSCDCControlModes/VSCACControlModes enums and promoted REMOT/RMPCT to first-class remote_bus_control_*/rmpct_* fields. PFFP keeps booleans and drops REMOT/RMPCT entirely. Different component model — but if you want VSC remote-regulation round-tripping, those two fields are simply absent today.

Divergences in behavior that look intentional:

• cb8ed3363 / 36a33acd0: PSY zeroes switched-shunt initial_status for MODSW ∈ (0,1,2) in all versions. PFFP reworked this — unconditional for pre-v35, actual S1..S8 for v35. PFFP's looks more correct; noting it because the behaviors genuinely differ for v35.
• 624516aec / cc34620d5 / 0cf823e18 / cca947511 / e7e8b6639: the branch_type_override / network-reduction feature set. PSY-only, no PFFP counterpart.

Strikethough on the first one: component behavior is tied up in this one, too. See PSY #1774: once those struct changes land, we should check back on the parsing situation here.