The existing logic of only updating the fee if it is not within 2x of
the current 2-block-eta does not work well for the current mempool.
The current mempool looks a bit weird: you need ~20 sat/vbyte to even get into it,
but there is only around 5 MB of txs paying >25 sat/vbyte.
The estimates look like this:
```
>>> config.fee_estimates
{144: 25764, 25: 27075, 10: 34538, 5: 34538, 2: 34538}
```
This commit changes the logic so that we send update_fee if the old rate is
- below 75% of the current 2-block-eta (instead of 50%), or
- below the 25-block-eta
To reproduce, open two wizards in parallel. Use one to enter the flow and start creating a wallet,
then physically unplug the hw device at the correct time, and use the other wizard to trigger a rescan.
The rescan will unpair the hw device, resulting in device_manager.client_by_id to return None when
continuing the flow on the first wizard.
fixes https://github.com/spesmilo/electrum/issues/8858
fixes https://github.com/spesmilo/electrum/issues/8853
This is a bit ugly, but activeFocus is declared as final and cannot be overridden directly...
```
1.42 | W | gui.qml.qeapp | file:///home/user/wspace/electrum/electrum/gui/qml/components/controls/ElTextArea.qml:23:5: Cannot override FINAL property
```
commit 43ab4a779a introduced "pyperclip" as a dependency for the prototype "text" gui.
pyperclip is small enough for us to vendor it, and looks mature/stable enough so that it would ~never need updating.
This makes it easier to try out the text gui when running from source, or using the AppImage.
> These kind of checks look incorrect and confused/confusing for me:
> 43a5d03426/electrum/gui/qml/qeinvoice.py (L388)
> 43a5d03426/electrum/gui/qt/main_window.py (L1801)
> For example, consider creating a local tx that spends all UTXOs in the wallet, and consolidates them to another ismine address.
> Such a tx basically does not change what wallet.get_balance() returns.
> but to the checks listed above, the confirmed balance of the wallet I guess should be 0?