Add custom clj-kondo hooks for macroses to ensure that lint rules are followed inside them.

Macros to be handled:

• fx/defn - similarly to defn with check on metadata for ::events
• fx/merge - seems like it could be lint'ed as thread macro, plus warn on limitations
• defview
• letsubs - lint as let + check on types for

GetVerboseBlockInfo uses NBitcoin's SendCommandAsync, which parses the returned JSON with Newtonsoft.

Then our RpcParser parses it again with Newtonsoft.

There're 2 opportunities to improve the performance significantly:

• Don't parse the JSON twice.
• Parse the JSON with System.Text.JSON instead.

The performance gain was about 80% when I messed around with it (https://github.

See https://github.com/lexansoft/ethcracker/tree/master/src/accounts/keystore/testdata

Add the ability to go above the year to see longer graphs.

Technical notes:

• that means asking more dates to countervalues API. Some changes needed in live-common. For a given currency we can infer the afterDay to use by using the older operation
• if oldest operation of the chart is under 365 days, the graph can fallback on the year mode
• it might be recommended switching to week

We avoid using cgo anywhere in the codebase since it causes problems with cross compilation. However, cgo would be advantageous in some cases (e.g. can use certain optimized libraries, can use libsecp256k1 bindings)

Investigate the viability of using xgo to do this. Test it on linux/osx/windows.

• https://github.com/karalabe/xgo
• https://github.com/ethereum/go-ethereum/wiki/Cross-compili

1. Start wallet from scratch
2. Go to Receive tab
3. Assuming Receiving address that got showed up first had no label, button would be Add Label
4. When you click on a Receiving address that is already labeled, button will change to Edit Label, and when you select again an address without label, button will stay Edit Label instead of Add Label.

![addLabelBug1](https://user-images

this is the natural behavior of node-hid but we should backport this behavior to all other transports.

basically it's a failsafe for race conditions.

if you open() a transport, you are responsible to close() it in a finally, if an open() happens during that time it should throw if it concerns the same device (e.g. by path id / by uuid, ... depending on the transport)