__UHEAP_FAILNEXT tests

It's calling SomeFunction (should really be called SomeFunctionL) with the heap failure rate increased once each time.  For example, with i set to 1, the first memory allocation SomeFunctionL does will fail (leave with KErrNoMemory).  The next iteration round, with i set to 2, the first allocation will succeeded, but the second will leave with KErrNoMemory.

It's all about testing the code in SomeFunction for cleanup handling  -  making sure evey object goes onto the cleanup stack, and in tern making sure everything gets cleaned up nicely if memory allocation failes.

For example, if SomeFunction was coded like this:

void SomeFunctionL()
{
HBufC* p1 = HBufC::NewLC();
HBufC* p2 = HBufC::NewLC();
HBufC* p3 = HBufC::NewL();
HBufC* p4 = HBufC::NewLC();

CleanupStack::PopAndDestroy(p1);
CleanupStack::PopAndDestroy(p2);
delete p3;
CleanupStack::PopAndDestroy(p4);
} 

Then the testing code would panic with an alloc panic when i = 4.  Why?  Because the allocation for "HBufC* p4 = HBufC::NewLC();" is set to fail, causing a leave with KErrNoMemory.  This will cause p1 and p2 to be popped and destroyed as expected, but will cause a leak from p3 since it's not on the CS when it should be.  If you didn't write testing code like above, you might never notice that bug, since it relys on the allocation for p4 failing.  If it doesnt, as it wont 99.9% of the time, the code will pass without error.