So it seems he used the distance between the supernova and the dust cloud (not the total distance). Then he divided by the time interval (measured on earth).By measuring the distance between the supernova and the surrounding gas, and dividing by the time between the two events, we can compute the speed of light when the supernova happened.
This time is affected by something else. He seems to implicitly assume light from both objects (cloud and supernova) takes the same amount of time to reach earth coming directly from each respective object; this is not valid for CDK. Due to CDK, the starting speed for light emitted later in time would be slower than for light emitted earlier in time. Thus, the light traveling along path B below is traveling slower, and takes longer, to reach the earth. This is assuming both cloud and supernova are equidistant from earth. The article does not state which is closer to earth, or if they are the same distance from earth.
Fig. 1 - Supernova 1987A & dust cloud
He assumes the time difference between the arrival of light from the two objects on earth, to be due to the delay caused by light traveling between the objects. But something else impacts this time; light traveling to earth, not between the objects, but from each object to earth, reaches earth at different rates. Light takes longer to reach earth if light travels slower, even over the same distance. So light from the cloud, even over the same distance, leaves later when c has decayed more, so is slower, and takes longer to reach earth. This adds to the time interval, while the faster light between the objects subtracts from the time interval. These 2 effects might cancel - they are in opposite directions so it is possible. At any rate, the calculation is not valid to disprove cdk.
So he seems to be dividing the distance between the objects by the difference in arrival time of light from those objects. This leaves out the fact impact of cdk on the time.
The time between arrival on earth is assumed to be due to the time for light to travel from supernova to cloud.
Consider the arrival time on earth for light from the dust cloud. Per CDK, light travels faster between supernova and cloud, so light from the cloud will get to earth faster. However, there is another effect to make the light reach earth later, which is not considered.
It is simple; the light from the cloud left later than light from supernova 1987a, so travels slower (since c had already decayed some). Thus light from the cloud travels the same distance to earth, but starts out slower. Thus it takes longer to for light from the cloud to travel the same distance to earth, compared to the light from the supernova.
This adds to the time difference between the two events measured on earth.
So, CDK results in adding time to the measured difference due to slower travel for light from cloud to earth, and cdk also results in reducing time due to faster-than-today's-c travel between supernova and cloud. These two effects of cdk tend to cancel each other - whether they cancel completely, or partially, requires more calculation.
To evaluate this, one needs the cdk equation for how c changes with time, the actual distance between the supernova an dust cloud, their distances to earth, and the precise times of arrival at earth of light from both cloud and supernova.
What was done it, seems, was a much simpler calculation, quoting Lisle:
... measuring the distance between the supernova and the surrounding gas, and dividing by the time between the two events, ...
Fig. 2