What is the buffer time between gates of the ibmq_essex in nanoseconds?

This is a question that I asked on Qiskit ignis section of slack, with a very interesting answer by A Kapila, but I also need to get a short answer in nanoseconds.

I want to evaluate decoherence in function of a waiting time introduced by adding identity gates before the measurement phase (using barriers before and after the series of id gates)

For instance for the Essex device, I can obtain the id gate length value using the following code:

device = provider.get_backend(‘ibmq_essex’)

properties = device.properties()

from qiskit.providers.aer.noise.device.parameters import gate_length_values

gate_times_out = gate_length_values(properties)

print(gate_times_out[0])

 
(‘id’, [0], 64.0)

Note that this method of creating a delay was used in a published paper about decoherence.

https://arxiv.org/pdf/1712.07080.pdf

In the methods section the authors give the following description:

« A delay τ is realized via the application of multiple identity operations. Each identity operation has the same duration as a single-qubit rotation gate (80 ns) and is followed by a 10-ns buffer time, which corresponds to a total delay of 90 ns per identity operation. »

For the circuit presented here, the waiting time is 4 * 64 ns = 256 ns plus four times the additional buffer time between gate for the device.

I can find the buffer times in older descriptions of some devices, like the Yorktown (see: https://github.com/Qiskit/ibmq-device-information/blob/master/backends/yorktown/V1/version_log.md).
But I cannot find it for the Essex device.What is it? Or where can I find it?
Or maybe the waiting time is already included in the gate length values?

Thanks by advance for helping.