Remove the previous epoch spent text files from IRI

[GalRogozinski]

Description

Make IRI independent of the previous epoch spent text files and reliant only on the external spent-addresses-db

Motivation

Those text files take lots of space in IRI and bloat up the jar. They are not needed if we require people to use an external db

Requirements

1. Delete the files from IRI
2. Don't check for their existence and don't try to load them

[jakubcech]

This assumes people don't start from a global snapshot (sync from scratch). And if they do, then they need to have a way of getting the spent addresses data - e.g., with the release.

[GalRogozinski]

Ok, I want to extend this issue now to extend the global-snapshot.

The new requirements are:

1. Create a new local-snapshots-db that has all the spent-addresses and global snapshot
2. Sign it
3. Upload it to an agreed place (TBD)
4. Remove the accompanying snapshotMainnet.txt and previousEpochSpentAddrersses.txt
5. delete logic for reading from files
6. change milestoneStart Index to 0 in config (but put the global snapshot index in a comment) please.
7. Repeat for devnet

Extra motivation:

Faster load time

[DyrellC]

We will need to have a conversation with devops to have them sign and post the spent-addresses-db to an s3 bucket or some form of IF based storage.