Move to the new hyperdrive-daemon (WIP)

[pfrazee]

This is a work-in-progress PR to replace the dat-daemon with the new hyperdrive-daemon which will also move us to the new Dat 2.0 stack. I will update the TODOs as this progresses.

TODOS

• Replace the daemon getArchiveInfo()
  • Need a way to fetch the latest version (Method to fetch the latest version of an archive hypercore-protocol/hyperdrive-daemon-client#5)
• Implement all existing hyperdrive methods
  • history (relies on upstream)
  • access (relies on upstream)
  • download (relies on upstream)
  • createReadStream
  • createWriteStream
    • Support for mtime, ctime params
  • createDiffStream
    • Need the .version of each change
  • watch
    • Missing path in event Include node path on watch() 'changed' event hypercore-protocol/hypertrie#24
  • createNetworkActivityStream
• PDA2
  • stat() .blocks and .downloaded (relies on upstream)
  • readFile opts (relies on upstream)
  • writeFile opts (relies on upstream)
  • mount() should resolve DNS before mounting
  • update docs
• Resolve whether the .key file should be part of drives
• Occasional "Error: Failed to connect before the deadline" on startup hypercore-protocol/hyperdrive-daemon-client#3
• Add fns to stat output (isFile, etc) hypercore-protocol/hyperdrive-daemon-client#4
• Update hyperdrive-daemon to level@5
• Issue: Mounting at a version is failing
• Issue: isSymbolicLink() is not returning true
• Issue: first visit needs to wait for data before attempting to read (getting 'no file found' because not waiting for a file listing to come down)
• Issue: Daemon will sometimes fail with "Bandwidth exhausted"
• Implement the mount API
• Implement the symlink API
• Implement network events and stats (relies on upstream)
• Implement disk-usage tracking (relies on upstream)
• Implement local disk-usage management including clearing all stored data (relies on upstream)
• Test folder-sync
• Move the exportArchiveToArchive algorithm out of the old daemon and into beaker (fork relies on this)
• Move the folder-sync algorithm out of the old daemon and into beaker
• Move the "preview mode" scoped-fs management out of the old daemon and into beaker
• Replace or deprecate the auto-download behavior
• Deprecate or replace the dat bandwidth throttle config
• Deprecate or replace all debug/logging in the old daemon
• Deprecate or replace the wire-protocol extensions (datPeers)
• Move the daemon into an external self-managing process (relies on upstream)
• Create a migration flow for previous user-data

[pfrazee]

Achievement unlocked: Loaded a hyperdrive 10 site (from Copenhagen!)

[andrewosh]

hyperdrive-daemon 0.10.0 has a few changes that should fix a few of your remaining issues:
New methods:

1. createDiffStream, which you can use as a replacement for history for the time being.
2. checkout, which behaves the same as Hyperdrive.checkout, except it is async and returns a RemoteHyperdrive. You can use these with createDiffStream using Hyperdrive's API.
3. version, an async method which returns the current drive version.
4. There are client/server implementations for download, but don't use that one quite yet (Hyperdrive impl still pending).

Bug fixes:

1. isSymbolicLink should return true for symlinks now, as stat returns the same object that's used in Hyperdrive (factored out into a separate module).
2. readFile accepts an encoding option now.
3. .key is no longer written as an actual file to any drives.
4. The connection timeout's been increased, and a connectionTimeout option has been exposed so you can tweak it as you need (this will need to be given more thought, too).