Change node handling of locations for persisting data

[Maelkum]

This PR changes the default locations where node persists files - the workspace, peer db and function db. Motivation for this is to ease the use of nodes when starting them directly. Before, if you'd start a head and worker node directly, you'd need to specify the paths directly else the nodes would trample each other's files (e.g. use --peer-db worker-peerdb --function-db worker-function-db ).

Changes are not completely backwards compatible.

We differentiate three scenarios.

1. User specifies workspace, peer db or function db paths directly via CLI flags - in this case, we use those

2. User starts the node and specifies a private key.
   In this case, it will store its files in .b7s_<peer-id> directory, in appropriate directories; for example:

   • .b7s_12D3KooWHUeKgXT4aj8oKvtwovVMki468igSsa5F8izZY3U5UyMD/workspace
   • .b7s_12D3KooWHUeKgXT4aj8oKvtwovVMki468igSsa5F8izZY3U5UyMD/peer-db
   • .b7s_12D3KooWHUeKgXT4aj8oKvtwovVMki468igSsa5F8izZY3U5UyMD/function-db

3. User starts the node without a private key
   In this case, node has a random identity. IMO, this is mostly used for testing and transient stuff anyways, so it's somewhat expected that the DBs and workspace is transient too. In this case, a new temporary directory will be created where files will be stored, e.g. /tmp/.b7s_2379210846.

I think these changes allow for more control over the node behavior and increase usability.

Also, fixed one bug that was introduced where we didn't detect if both runtime-cli and runtime-path CLI arguments were empty. Such a worker node would fail all executions and we should detect such an invalid configuration on boot.

WDYT?

[dmikey]

I like this rearrangement. Makes it much easier to run multiple nodes on bare env.