Scan save mem rewrite masks issues with `steps=0`

Description

import pytensor
import pytensor.tensor as pt
from pytensor.compile.mode import get_default_mode

n = pt.iscalar("n")
x0 = pt.vector("x0")
xs, _ = pytensor.scan(lambda xtm1: xtm1 + 1, outputs_info=[x0], n_steps=n)

out = xs[-1]  # Invalid when nsteps=0

fn = pytensor.function([n, x0], out)
print(fn(n=0, x0=[0, 1]))  # [1. 2.]

fn = pytensor.function([n, x0], out, mode=get_default_mode().excluding("shape_unsafe"))
print(fn(n=0, x0=[0, 1]))  # [1. 2.]

fn = pytensor.function([n, x0], out, mode=get_default_mode().excluding("scan_save_mem"))
print(fn(n=0, x0=[0, 1]))  # IndexError: index out of bounds

I suspect from this hack: https://github.com/pymc-devs/pytensor/blob/8454c3ba78b1889987a909745f87c8dcab8e48fb/pytensor/scan/rewriting.py#L1438-L1445

But removing this hack leads to some tests failing, so other stuff may be doing wrong assumptions downstream of it (or perhaps inside the rewrite)

Start with the reproducer in the issue and inspect pytensor/scan/rewriting.py around lines 1438-1445. Compare behavior with the default mode, shape_unsafe excluded, and scan_save_mem excluded, then run the tests that fail when the hack is removed. Done means steps=0 no longer produces an incorrect result while the existing scan rewrite tests continue to pass.