With 30m between your battery/power shed and caravan, you really need an isolated interface to avoid the risk of nearby lightning strikes inducing large enough transient ground potential differences or EMI induced transient currents in the signal cable to blow the s--t out of your monitoring and control devices. Although its possible to use isolated RS485, I'd recommend a CAT5 or better Ethernet cable with surge protectors directly tied to local ground and an Ethernet shield for the Arduino in the shed, as Ethernet inherently has >1KV isolation at each end, provided you do *NOT* use STP patch-leads between the surge protectors and the devices they are connected to (so you aren't creating ground loop by linking the device chassis to the ground to the surge protector). Once you've got Ethernet, any extra monitoring, command and control you need in the power/battery shed is easy to add over the same cable. What you then use to display the data in the caravan (or later on in your house) is up to you - e.g. web browser or program on the PC, app on your phone or tablet, a hacked ebook reader or a dedicated custom built ethernet connected display.
Whether you use high side or low side current shunts is irrelevant so long as the instrumentation amplifiers, or other types of readout circuit can handle the maximum possible common mode voltage that may be present during fault conditions without damage. This could easily reach double the solar panel string's maximum open circuit voltage due to cable inductance ringing with power input decoupling capacitance of various modules in your system when breakers are initially closed. Your proposed INA240A3 only has an 90V abs. max. common mode voltage rating, (and the INA238 Renate suggested only 85V) which even for a low side shunt is inadequate if a bad connection develops between the ground side shunt load terminal and DC system ground. Depending on *exactly* where in the circuit the low side shunt is used, and what's loading the panel, its sense terminals could get driven negative by as much as the panel open circuit voltage, + whatever transient spikes may occur. See mitigation suggestions below.
Your alternative of an ADS1115 doesn't offer any common mode input voltage range outside its supply rails so it and any Arduino its connected to are unlikely to survive any open ground faults if its directly connected to the shunts without extra input protection. One way to mitigate that would be to minimize the risk of an open ground fault between the two shunts, and the ADS1115 and Arduino by bolting the shunts to a heavy copper ground busbar with separate bolted connections for the battery negative cable and the Arduino + ADS1115 ground to minimize the risk of the Arduino and shunt common connection being disturbed if the battery negative cable is moved or disconnected. Use a telecom* rated nom. 48V input, 9V output DC-DC converter to power the Arduino via its Vin pin or one with a precision regulated 5V output to power it via its +5V pin. During development, using an USB isolator is *STRONGLY* recommended so you don't blow your PC's USB port.
* Telecom applications historically had large nom. 48V lead acid battery banks, positive terminal grounded to provide an approx -55V (float voltage) supply for exchange equipment and individual subscriber loops, and that power voltage remained an industry standard when the industry went digital, so rugged power conversion solutions for it are readily available.