Guardrails are not easy to understand

Issue in ./help/ingestion/guardrails.md

One example, for the below guardrails,

Batch ingestion to Profile | You can ingest up to 120 GB of data per hour.The maximum size of a record class is 100 KB (soft).The maximum size of an ExperienceEvent class is 10 KB (soft).The maximum size of a single record is 1 MB.
-- | --

Customers are not able to understand,

1. What is a record class, this should be given as a clear reference to profile class or custom classes.
2. When we say 100 kb or 10 kb, how is this computed, is this just a record in single line json or something else?
3. What happens when a soft limit is breached, can we describe it.
4. What happens when a hard limit is breached, can we describe that.
5. How is a batch created when we are not loading data via a file i.e. lets say we use query service to transform data into a dataset which is enabled for profile. How big will be the batch size then?

Start with help/ingestion/guardrails.md and review the existing guardrails table and terminology. Determine how record classes, record sizes, soft and hard limits, and batch creation are defined, then update the page so each point is explained clearly and the listed ingestion scenarios are addressed.

Written by the indexing model from the issue text.