    BOB is having problem with using the clock sentinel.  Every time BOB starts anew, I have to take it to a window to get satellite time.  I tried erasing the clock.txt file and the directory, and a new directory and file were created, but it does not rely upon the clock.txt and forces a satellite time reset.  I tried clicking RESET on BOB, and it needs satellites.  RESET on the three others does not, they resume without needing satellites.
    
    Also, the splash screen with version needs:
    1) the version needes to be on a 2nd line, it goes off to the right of the screen, so i cannot determine its full version.
    2) the splash screen needs to hold for 10 seconds, it flashes by too quickly, so please introduce a sleep.
    
    ED never seems to link.
    
    I changed in main.cpp:
    
	    //static constexpr uint32_t LINK_RETRY_MS = 30000;
	    static constexpr uint32_t LINK_RETRY_MS = 90000;

because the LINKS were timing out.  Still, with 90 seconds, they time out.

Is the continued ANNOUNCEMENT something that is interfering with the longevity of LINKS?  It seems that if a LINK is established, then both units on the LINK would just continue sending back and forth; that is not happening, the LINK seems to be torn down.

I have a 3 minute set of logs with the lastest code at: ~/logs/20260528_1707.  Of the four 180 second logs, only one entry of:

	jlpoole@jp ~/logs/20260528_1707 $ grep -i 'RX LINK' *
	CY_raw_20260528_170755.log:RX LINK: inbound link established hash=3255588dd68ff2bc6ab95b8712120343
	jlpoole@jp ~/logs/20260528_1707 $ 

Are we hitting some bugs in microreticulum?

When you are ready to compile, please compile only 1 unit, if that compiles, then so will all the rest.  Notify me and I can run a bash script that will compile the others -- it takes about 4 minutes to complete.

Please leave all uploading of newly compiled firmware to me.

I am trying to maximize your time within my 5 Hour allowance and which recently has been consumed at a rapid rate and I think having you monitor units and wait for compiles and upload may be eating into the budget.  I've developed a logging system that isolates logs from a run into its own directory, e.g. ~/logs/20260528_1707, so we can use that to efficiently evaluate what is going on.