Corner Tightening in Adaptive body bias

Can anyone explain what is Corner Tightening in body bias? I have some confusions regarding this topic. Please explain

Sample links where the idea is mentioned:

semiengineering.com/the-return-of-body-biasing/

soiconsortium.org/2019/02/20/gf-dol...-energy-savings-faster-ttm-design-kits-q2-19/

www.globalfoundries.com/news-events...hin-integration-deliver-differentiated-fd-soi

You'd like your body biasing to end up making
some device attribute as-desired. Without making
a mess of another.

Global bulk biasing in FDSOI can (for example)
make PMOS leak while driving down leakage on
NMOS.

You can put a body-biased device inside a feedback
loop against some leakage proxy, and make the
NMOS shut up.

You could create a feedback element which looks
at minimizing the sum of NMOS and PMOS leakage,
to put you in Happy Valley.

Putting the body parasitic BJT into forward active
operation, or even snuggling up to it, is not great
for some environments where you'd like a hard
low-Z body tie (even if you have to eat some surface
channel leakage). And this can be a failing of servo
body bias schemes, an insufficient drive and response
time to deal with devices that are being pushed
toward snapback / single transistor latching / 4-layer
latchup / impact ionization / breakdown (and using
a device at one Vds to control others at different
Vds and Vgs, might be sloppy).

But why corner tightening is required ? is it because we need optimised performance and reduced delay ?