Update Documentation to Note to import before sample call

[davage]

In order for the inspector show function to work properly, the import statement must be called before the sampler sample function call, followed by which the inspector show function can be called.
i.e.:
import dwave.inspector
...
sampler.sample(Q)
...
dwave.inspector.show(Q, response)

This is not entirely obvious based on context and current documentation.

[randomir]

That is not entirely correct, since you do not have to activate the full data capture (done on import) to use show(), but it certainly makes the usage easier.

Data captured includes the problem sent to the QPU (including parameters) and the samples returned. It also captures warnings generated during embedding, and the embedding context (the map, and the chain strength).

So for completeness, the alternative to:

import dwave.inspector
sampleset = sampler.sample(bqm)
dwave.inspector.show(sampleset)

looks something like:

sampleset = sampler.sample(bqm, return_embedding=True, warnings='SAVE')
import dwave.inspector
dwave.inspector.show(bqm, sampleset, sampler)

To be clear, data capture is encouraged, because going forward it'll include much more data (like nested embeddings #40, but also other problem transformations), so we want to emphasize it our docs.