Instead of trying to find faulty keys, we stick
to the mechanics we have in place. This way,
we have less false positives.

However, we enforce a certain Bisq version
format, as parsing and business logic relies
on that now.

Turns out, scanning resources does not work reliable enough. Thus,
an array of Strings denoting historical Bisq versions it is.

An other way to put it is it is an array denoting which data stores
are there in the resources.

If we use the diff sync between seed nodes we create a race condition
where when a seednode gets updated to the new system, it does not
sync up properly with the other seed nodes. And that would be fatal.

So for the time being, when a seednode asks for data, it uses the
"old" big requests with all object keys. Should not be a problem
for now since they have enough bandwidth.

CI does have troubles with tests which do file operations. Thus,
these tests have been disabled. They have been useful during
development and are useful for testing locally, though.

This test addresses the migration scenario where a user does not
upgrade on the first possible occation (to the first Bisq version
that has the new database structure in place) but does so later.

Previously, we only checked if the newest store is there. If it is
not, we copy and reload everything. If it is there, we are satisfied.

Now, we check for each and every store we should have and copy and
load them individually.

Up until now, the tests needed to create a version history set for testing.
These changes had to be made to a final static variable during runtime.
Now, enough versions are in the history set anyways, so we could remove
this part of the testing framework.

I thought I could get by using a fixed version string so we can
have tests involing future data stores...

# Conflicts:
#	core/src/main/java/bisq/core/support/dispute/agent/MultipleHolderNameDetection.java
#	core/src/main/java/bisq/core/trade/txproof/xmr/XmrTxProofService.java
#	p2p/src/main/resources/AccountAgeWitnessStore_BTC_MAINNET
#	p2p/src/main/resources/SignedWitnessStore_BTC_MAINNET
#	p2p/src/main/resources/TradeStatistics2Store_BTC_MAINNET

storage.initAndGetPersisted and initAndGetPersistedWithFileName are not needed.

Refactorings:
- Extract variables for better readability
- Rename variables
- Split copyAndSplit to getStoreFromResource and pruneStore which describes better what is happening
- Use getStore from base class, remove local readStore method

Avoid that we have to iterate all historical stores at each getMap call and create that map once when the historical data are created.
Reduce visibility
Make fields final (readFromResources is not called twice)
Rename variable for better readability

We should not risk to break those critical data and it does not provide that much benefit as with the trade statistics. Lets test it first with trade stats and if all goes well extend in a later release.

… AppendOnlyDataStoreService to call the new method only where it makes sense.

Remove the hack of adding a "magic" key for communicating the version. Use a new field in the request instead.

Remove the extra handling of seedNode as the impl. did not make any difference  as the Version passed in
`excludedKeys = getKeySetInBytes(this.appendOnlyDataStoreService.getMap("since " + Version.VERSION));` was same as in
`if (!finalFilter.equals(Version.VERSION)) ` thus never doing anything there and the result maps is the same as in the seed node branch.

Was not clear anyway what the motivation of a diff. handling is there? That seed nodes request always all data without filtering? Why?

…re all historical immutable data.

Add ignoreHistoricalData param to getMap

Remove unused variables/params

The store has nothing to do with the splitting of live data and historical data.
It only provides a getMap method with the PersistableNetworkPayload type

Use that for all stores with such a type of map.

…ersion compare.

…historical data store feature.

@freimair Feel free to adjust the test to the new code.

Some tests are still failing. Need to look into it another day...

# Conflicts:
#	build.gradle
#	desktop/package/linux/Dockerfile
#	desktop/package/linux/package.sh
#	desktop/package/linux/release.sh
#	desktop/package/macosx/create_app.sh
#	desktop/package/macosx/finalize.sh
#	desktop/package/windows/package.bat
#	desktop/package/windows/release.bat
#	relay/src/main/resources/version.txt

…m/chimp1984/bisq into chimp-reduce_initial_request_size

AppendOnlyDataStoreService does not have a map but aggregates the data form its services.

We use a set at the filter so it is not deterministic which item get truncated.

If we get logs from windows users the missing line breaks makes it harder to read.

…-versioned resource file anymore. Instead use readStore(); to create the live store.

Add more logs

…esources.

Improve logs

# Conflicts:
#	core/src/main/java/bisq/core/trade/statistics/TradeStatistics2Store.java
#	p2p/src/main/java/bisq/network/p2p/network/Connection.java