However, for information, the configuration continues to work and the CPU problem has been relatively resolved...
In the end, there was nothing wrong with the routing rules.
It was the hardware that was not working properly. The device on which everything was running was a hEX POE (single core at 750 MHz, I think)...
I moved everything to one of the netPower devices that manages the first tower, and now CPU usage is around 40-50%.
On a network that size put BGP between all the sites and PROFIT.
You are trying to do manually what BGP will do automatically.
The best bit is because you have it all running on statics you can bring the BGP up site by site leaving the statics as backup routes and you use the statics as the BGP transit IPs. When you bring the BGP between sites the distance will be higher than the statics so your statics will still take precedence. When you are happy the BGP routes are good just take the distance up on the statics and the BGP will take over.
So in short you don't have to change anything significant just add the BGP over your current network.
Its rather trivial to setup lets take site A to Site B in your drawing
You have a transit link with an IP network between them you are statically routing I will use 10.66.66.1/30 and 10.66.2/30 to represent it. Let assume you have local network 10.0.1.0/24 at site A and 10.0.2.0/24 at site B
So substitute your transit IPs and local networks but it looks like this
SITE A 10.66.66.1/30 <==============> 10.66.66.2/30 SITE B
local network 10.0.1.0/24 …………………………… local network 10.0.2.0/24
The iBGP between them using a local public AS number lets say 65534 and lets set route distance to 10