My comments are based on mag particle testing a bunch of engine parts when working in a specialty machine shop in the last millenium.
=================.========
I'd say down at the Crank shaft end, keeping the big end bore round at the exhaust stroke TDC is very difficult, and any geometry anomalies ( rod bolt head clearance notches etc) are likely to be the highest stressed areas ans subject to fatigue failures.
Fatigue failures of the ubiquitous I=beam shank were pretty rare, and likely to originate at a major surface defect, often man made, like filing or sawing notches in the I-beam to number the con rod. The orienation of the I-beam shape is doubtless convenient for forging a production part, but if the section properties needed more beef in the fore and aft direction, then they would have been forged as a different shape I believe.
Up by the wrist pin end failures were pretty rare indeed.
===========.
Fred Carillo was a proponent of H-beam rods, but those may have had an advantage for machining from billets for custom lengths.
Their superb metallurgy and quality may have more do to with Carillo rods reputation for durablity as the choice of beam section choice.
============.
Under extreme boost ( and perhaps even more extreme detonation) or heavy doses of nitro or hydraulic locking or when valve heads break off and jam between the piston and cylinder head, when compression loading dominates and can be uniform or highly eccentric, almost anything can happen.
===============.
In the days before the availablity of high performance connecting rods, modifying stock con rods by various means was about the only choice.
Carefully done "Boxing" was a means to buy short term reliability.
Mag particle inspection would often reveal problems developing in the welding, as might be expected.
