First of all, for what you describe (distinct dst-nat rules per public IP), you do not need to create a macvlan interface, it is enough to attach both public addresses to ether1 and simply let some of your dst-nat rules match on
dst-address=x.x.50.24 and some on
dst-address=x.x.50.25.
You only really need macvlan in more complex scenarios like VRF, where it is indeed necessary that packets that came to a particular address on a physical interface are treated as if they came in via some other interface. There is nothing bad about using a macvlan interface in general, but the ISP may not be ready to handle its made up MAC address, whilst the use of an individual MAC address is the key component of the macvlan concept.
To make it clear whether the ISP has a problem with that unexpected MAC address or not, enable the macvlan interface (during a time when you can ignore the fact that it causes the internet connection to break), run
/tool/sniffer/quick ip-address=x.x.50.25 port=4567 and try to open
http://x.x.50.25:4567 in a browser on your mobile connected to LTE or 5G network; if the sniffer shows packets to come in, you can probably keep the macvlan in use, otherwise you'll have to move the x.x.50.25/24 to ether1 and try again. But the ISP may still have an issue with outgoing packets with the macvlan MAC address as source, so the above is only a first step check; the next step would be to ping the x.x.50.25 from the phone (which requires an app).
Other than that, you have set
use-ip-firewall under
/interface bridge settings to
yes; I suspect you have done that due to a misunderstanding of the purpose of this setting as I cannot see any need for setting it to
yes in the rest of your configuration. Setting
use-ip-firewall to
yes messes up with NAT quite a lot, so unless you can explain why you need it, set it back to tde default
no.