Emit an event to state approximate record size, to check for reaching 400KB record sizes

[a-h]

Consider whether it's possible to implement a pointer system.

For example, given the input data:

{
  "a": "large_blob",
  "b": "small datum", 
}

The library could be designed to break it up into pointers to the large data, i.e. would store two records:

The state where a key is replaced with a pointer.

{
  "a": { "stream_pointer": "hashOfBlob" },
  "b": "small datum"
}

And a pointer record:

{
   "_pk": "hashOfBlob",
  "data": "large_blob"
}

When the data is requested, the pointers would be followed to collect the data from the table.