Switch to stellar-core run-from command when added to Stellar-Core

[bartekn]

In stellar/stellar-core#2634, we proposed to add a new command to stellar-core: run-from [starting_ledger] that would start streaming from starting_ledger to the latest closed by a network and then would continue streaming new ledgers as they are closed.

The current workaround we use when starting a new captive session is the following chain of commands that use a persistent DB:

• stellar-core new-db,
• stellar-core catchup [starting_ledger-1]/0,
• stellar-core run.

While it works fine, it has some disadvantages:

• It requires specifying a DB connection and maintaining a DB.
• Applying buckets stage is much slower when using a disk vs RAM.
• stellar-core run execution can be slow when it needs to wait for the next checkpoint.

The new command would solve all the issues above.

[bartekn]

Fixed in #3163.