Don't require passing --signer.dir when generating/signing transactions with ledger signer

SUMMARY

When using a Ledger device to sign transactions, currently one needs to first export entity's id (public key) for the given account index via:

oasis-node signer export \
  --signer.backend ledger \
  --signer.dir /localhostdir/entity/ \
  --signer.ledger.index 1

and then specify the path to entity's id by passing --signer.dir flag when generating/signing transactions, e.g.:

oasis-node stake account gen_transfer \
  --genesis.file /localhostdir/genesis.json \
  --signer ledger \
  --signer.ledger.index 1 \
  --signer.dir /localhostdir/entity/ \
  --stake.amount <AMOUNT-TO-TRANSFER> \
  --stake.transfer.destination <DESTINATION-ACCOUNT-ADDRESS> \
  --transaction.file tx_transfer.json \
  --transaction.nonce 1 \
  --transaction.fee.gas 1000 \
  --transaction.fee.amount 2000

To improve the Ledger transaction generation/signing process, a user shouldn't need to export his entity's id and pass it via --signer.dir flag.

ISSUE TYPE

• Feature Idea

Start with the oasis-node signer export and stake account gen_transfer command paths described in the issue, along with the linked Ledger documentation. Trace how the Ledger account index and entity ID are currently supplied, then determine how generation and signing can work without exporting an entity ID or passing --signer.dir. Done means the documented Ledger transaction workflow no longer requires that flag.