AD8210 input pins source current to power rail being monitored

This question has been posed no less than 3 times on this forum with no definitive answers from a representative of AD.

AD8210 sourcing current when input are floating - Q&A - Differential Amplifiers and ADC Drivers - EngineerZone
AD8210 sourcing current when high side switch is off - Q&A - Differential Amplifiers and ADC Drivers - EngineerZone
AD8210 - Vs bleeding thru differential inputs? - Q&A - Operational Amplifiers - EngineerZone

Problem statement:
When used as a high side shunt monitor with high side switches, the shunt inputs of the AD8210 (-IN and +IN) act as current sources allowing an unwanted voltage to appear on power line which the shunt is monitoring.
This issue does not exist with the TI INA240 series of parts, which have the same function and footprint (and are available in different gain options)
This effect is made particularly obvious with the use of indicator LEDs on the power rails which are being leaked to as they will appear slightly on when the power rail is meant to be off.

Replication steps:
1. Build the following typical application circuit.

2. Supply the AD8210 with 5V
3. Leave I_SENSE_IN and I_SENSE_OUT floating.
4. Watch a voltage appear at I_SENSE_IN and I_SENSE_OUT
5. Short I_SENSE_OUT to ground while keeping I_SENSE_IN floating
6. Observe ~2.5mA flow from the I_SENSE_OUT to ground.
7. Observe the 5V supply current increase by the leaked amount.
8. Observe the leakage current to scale with the supply voltage of the AD8210

My questions:
Is this leakage a known and intended behavior of the AD8210?
Are there any mitigations for this effect? i.e How do I stop the voltage rail I am trying to monitor from appearing at 3.2V when its disabled.
Can the functional schematic of AD8210 be updated in the datasheet to reflect the source of this leakage, for instance: