[Investigate] Can only utilise ~70% of scope data size limit.

Given a scope limit of 10 million bytes, I was only able to utilise ~70% of the limit using cbworkloadgen.

Logs: https://cb-engineering.s3.amazonaws.com/MB-49845/collectinfo-2021-12-01T100309-ns_1%40127.0.0.1.zip

Requirements:
A single node with a single 'default' bucket.

Steps to reproduce:
1. Create scope 'my_scope' with a limit of 10000000 bytes:

curl -u Administrator:password http://10.112.210.101:8091/pools/default/buckets/default/scopes -d 'name=my_scope&limits={"kv": {"data_size":10000000}}'

2. Create single collection inside the scope using the UI.

3. Use cbworkloadgen to produce 7000 items of 1000 bytes with the --no-compress flag set.

/opt/couchbase/bin/cbworkloadgen -b default -u Administrator -p password -s 1000 -i 7000 -c 0x8 --low-compression

What happens?

Cbworkloadgen fails with a 'MCSink MC error: 52'.

2021-12-01 09:53:23,586: s0 error: async operation: error: MCSink MC error: 52 on sink: http://127.0.0.1:8091(default@N/A-0)

error: MCSink MC error: 52

The 'sum(kv_collection_data_size_bytes)' statistic is at 6797930 bytes following this:

From the user interface I can observe that I have 6.44K items with a total of 6.48MiB disk utilisation for that particular scope:

What I expected to happen:

I expected to able to write ~10000000 bytes of documents to the scope.