Non defined modbus slave id's can be picked for devices

[sfstar]

Hello and thank you for opening the code to the modbus server component of the victron ecosystem.
There seems to be a mismatch between the slave id's picked for devices connected to the GX devices and the documented possible slave id's.

The XLSX states:

However it seems that other id's might also be selected for the PVinverter devices that are connected to the GX device.
In the issue on my issue tracker a user experencied the GX device picking 34,35,36 as modbus slave id's for his pvinverter devices.

For reference the issue on my integrations issue tracker:
sfstar/hass-victron#36

According to the XLSX data none of these ID's are valid pvinverter slave id's.
Is this by design?
If yes, why are slave id's 35 and 36 not defined in the XLSX?

My integration treats every slave id the same and attempts to retrieve all possible registers for each slave id.
So it doesn't matter to me if the devices don't adhere to the assigned unit id's (although it's still wierd that it diverges from the documentation).
However, I'm not testing id's that aren't defined in the XLSX.
Therefore, I'm also asking if you could provide me with an complete list of possible slave ID's so that i can reflect this in my integration.

[sfstar]

Relates to #36

[sfstar]

@izak Would you be able to provide me with an list containing all valid unit id's that the GX device can assign?