Not yet sure when and if it would be added. But you can do the same thing manually though even today. Though, I must note that after writing this answer this is somewhat clunky to do manually. Though, while the autoConvert is an ok starting point, I would say it would probably make more sense to sanitize the inputs first and report erroneous (e.g. overflows of numeric types from String) to the user before even calling the write. If it is instead some automatic process or somewhere you cannot show live feedback (e.g. as editing a text box in UIs etc.) it probably doesn’t matter that much.

Also, in the case of a single write operation the flow is trivial: if the conversion fails, just throw DataTypeConversionException and do not Write. However, in the case of multiple it is no longer so (should it succeed partially as OPC UA otherwise does have partial success as a concept? should errors be reported in the ‘statuscodes’ or a separate construct as it could be confused as a server error, but this complicates the return vs. current writeValues… etc.).

If you have the UaNode at hand you can check if it is UaVariable and cast and call getDataTypeId() for it. You can also cast it to UaVariableImpl and call writeValue(Object value, boolean autoConvert) directly to it (since the node does know the datatype already). There is no way yet to “compose” this Write from multiple nodes at once.
If you do not have it as UaNode (and wish not to get it due to perf reasons), you can Read the DataType Attribute (the value is a NodeId within the DataValue/Variant), this is basically what the UaClient does if the autoConvert is true i.e. it will first read the DataType Attribute (each time it is called). For other attributes than Value it uses an internal map for the Attribute type. The actual conversion is done via client.getAddressSpace().getDataTypeConverter().convert(Variant value, NodeId dataTypeId).

Thus you could e.g. do one Read for the DataType Attribute for all your nodes, then convert everything and then make one Write. Though, I think you will need to use the most generic UaClient.read(Double maxAge, TimestampsToReturn timestampsToReturn, ReadValueId… nodesToRead) for the Read part since we lack a helper method for a case of reading a single Attribute for multiple nodes (we do have the other way around of reading multiple attributes for a single node). We … might add that, though there is already so many different UaClient.read methods so it could also just be confusing it … more, thus I’m not like sure do we add it.