[clang] [analyzer] Add option assume-one-iteration (PR #125494)

[Donát Nagy via cfe-commits]

================
@@ -294,6 +294,16 @@ ANALYZER_OPTION(
     bool, ShouldUnrollLoops, "unroll-loops",
     "Whether the analysis should try to unroll loops with known bounds.", false)
 
+ANALYZER_OPTION(
+    bool, ShouldAssumeOneIteration, "assume-one-iteration",
+    "Whether the analyzer should always assume at least one iteration in "
+    "loops where the loop condition is opaque (i.e. the analyzer cannot "
+    "determine if it's true or false). Setting this to true eliminates some "
+    "false positives (where e.g. a structure is nonempty, but the analyzer "
+    "does not notice this); but it also eliminates some true positives (e.g. "
+    "cases where a structure can be empty and this causes buggy behavior).",
----------------
NagyDonat wrote:

> It could be mentined that this option interacts with analyzer option 'eagerly-assume'.

It's just a bug -- not an undocumented feature :upside_down_face: . 

The _implementation_ of my change needs to heavily interact with `eagerly-assume` but I'm trying to reimplement the same behavior with and without it. There are some contrived corner cases where enabling or disabling `eagerly-assume` changes the behavior of this `assume-one-iteration` logic, but I wasn't able to construct an example that could plausibly appear in real-world code, so I don't think that we need to document these theoretical buggy cases.

https://github.com/llvm/llvm-project/pull/125494