Older miniupnpd 2.0 will send a SSDP search response with an IPv6 location to a IPv4 address,
but newer ones 2.2 won't. So we need to also bind to an IPv6 address for the SSDP search
to receive the router's IPv6 location. Then we can bind to our public IPv6 address
for a port forward and it will work when miniupnpd is configured for "secure".

Also, don't bind a POST request to a mismatched v4/v6 address.

and make package private, so all the tunnel build classes are in one place.

by making it always nonblocking.
as reported by drzed

of peers. For efficiency.
ref: http://zzz.i2p/topics/3082
thx: jogger

- Generate and parse short record format
- Encrypt and decrypt short records
- Register handlers for 3 new messages ITBM/STBM/OTBRM
- Send ITBM/STBM if all hops support it (disabled)
- Reply with OTBRM at OBEP if STBM received (disabled)
- Send STBM at IBGW if ITBM received (disabled)
- Add logic for when to send new messages
- ChaCha encryption of other short records
- Fix compare logic in ITBM parser (ticket #2814)

All is still preliminary, disabled, untested; proposal is still incomplete

Still todo:
- Fill in plaintext record for ITBM/OTBRM
- OTBRM key/tag

Fix padding on file chooser buttons

Only check for clients.config if there was no clients.config.d

More test mods
Fix StringBuilderWriter.write(int)

Jetty server high CPU when client send data length > 17408

This affects SSL connections only, which is not part of our default setup.
Adapted from workaround at:
https://github.com/eclipse/jetty.project/security/advisories/GHSA-26vr-8j45-3r4w
Put the new checks directly in the unwrap() method,
rather than subclassing SslConnection, as that would require config file changes.

When checking whether to use Roaming appdata(like on older installs or on installs where people deliberately use Roaming appdata) also check whether clients.config exists and if it does, set home to Roaming appdata since people may be updating from routers before the migration to clients.config.d directories. This should fix #16

TODO: set outbound addresses on transition to firewalled later

for local interfaces even when not force-firewalled, because REA2()
now needs them to set the IPs regardless

Timer was not getting rescheduled in all cases,
so testComplete() was never called,
so no more tests could run.
Always remove test when testComplete() is called.
Log tweaks

for embedded applications.

Ensure a migration failure is logged if file does exist.
Reported and tested by zlatinb

Ensure introducers are required when force-firewalled
Log tweaks

Same as CDBQ earlier. To allow experiments on adjusted params.

Fix "very close" message when configured for all routers

it's already called from IMF

Log tweaks