You cannot embed the way you describe, but using Azure tables is optional. In the above example, I would set Item1 = Partition Key. Elements 2 → n will be a Row Key. Each row can have a unique schema, and each element will be part of the index, so the search will be very fast.
The row keys must be unique within the section, so if necessary, you can have a repeating pattern (it's hard to say from the description).
You can then get all the items associated with Item1 by querying PK or get a single item by specifying PK + RK.
source share