Loosening regex on phi merge zero generation regression tests

The tests that should have caught this were too specific when checking for the names of the variables and at some point in the past dynamic_object got given a prefix, which meant that the check to make sure no zero generation existed always passed (even with a #0 there) because the variable name wasn't exact.

Made the matching far more forgiving and it now looks for anything that vaguely looks like a guard statement with a 0 generation object.

Author: JohnDumbell

jbmc/regression/jbmc/phi-merge_uninitialized_values/field.desc

@@ -5,8 +5,8 @@ activate-multi-line-match
 ^EXIT=0$
 ^SIGNAL=0$
 --
-^.*:\s+(dynamic_object|new_tmp)[0-9]+(@[0-9]+)?#0\)
-^.*\?\s+(dynamic_object|new_tmp)[0-9]+(@[0-9]+)?#0\s+:
+\([^\s]+guard[^\s]+ \? [^\s]+#0 : [^\s]+\)
+\([^\s]+guard[^\s]+ \? [^\s]+ : [^\s]+#0\)
 --
 These regexes are making sure that a variable of generation 0 dosen't appear in a phi merge, so the below
 statement:

jbmc/regression/jbmc/phi-merge_uninitialized_values/local.desc

@@ -5,8 +5,8 @@ activate-multi-line-match
 ^EXIT=0$
 ^SIGNAL=0$
 --
-^.*:\s+(dynamic_object|new_tmp)[0-9]+(@[0-9]+)?#0\)
-^.*\?\s+(dynamic_object|new_tmp)[0-9]+(@[0-9]+)?#0\s+:
+\([^\s]+guard[^\s]+ \? [^\s]+#0 : [^\s]+\)
+\([^\s]+guard[^\s]+ \? [^\s]+ : [^\s]+#0\)
 --
 These regexes are making sure that a variable of generation 0 dosen't appear in a phi merge, so the below
 statement:

jbmc/regression/jbmc/phi-merge_uninitialized_values/static_field.desc

@@ -5,8 +5,8 @@ activate-multi-line-match
 ^EXIT=0$
 ^SIGNAL=0$
 --
-^.*:\s+(dynamic_object|new_tmp)[0-9]+(@[0-9]+)?#0\)
-^.*\?\s+(dynamic_object|new_tmp)[0-9]+(@[0-9]+)?#0\s+:
+\([^\s]+guard[^\s]+ \? [^\s]+#0 : [^\s]+\)
+\([^\s]+guard[^\s]+ \? [^\s]+ : [^\s]+#0\)
 --
 These regexes are making sure that a variable of generation 0 dosen't appear in a phi merge, so the below
 statement:

regression/cbmc/phi-merge_uninitialized_values/dynamic.desc

@@ -5,8 +5,8 @@ activate-multi-line-match
 ^EXIT=0$
 ^SIGNAL=0$
 --
-^.*:\s+dynamic_object[0-9]+(@[0-9]+)?#0\)
-^.*\?\s+dynamic_object[0-9]+(@[0-9]+)?#0\s+:
+\([^\s]+guard[^\s]+ \? [^\s]+#0 : [^\s]+\)
+\([^\s]+guard[^\s]+ \? [^\s]+ : [^\s]+#0\)
 --
 These regexes are making sure that a variable of generation 0 dosen't appear in a phi merge, so the below
 statement:

regression/cbmc/phi-merge_uninitialized_values/global.desc

@@ -5,8 +5,8 @@ activate-multi-line-match
 ^EXIT=0$
 ^SIGNAL=0$
 --
-^.*:\s+global(@[0-9]+)?#0\)
-^.*\?\s+global(@[0-9]+)?#0\s+:
+\([^\s]+guard[^\s]+ \? [^\s]+#0 : [^\s]+\)
+\([^\s]+guard[^\s]+ \? [^\s]+ : [^\s]+#0\)
 --
 These regexes are making sure that a variable of generation 0 dosen't appear in a phi merge, so the below
 statement:

regression/cbmc/phi-merge_uninitialized_values/local.desc

@@ -5,8 +5,8 @@ activate-multi-line-match
 ^EXIT=0$
 ^SIGNAL=0$
 --
-^.*:\s+local(@[0-9]+)?#0\)
-^.*\?\s+local(@[0-9]+)?#0\s+:
+\([^\s]+guard[^\s]+ \? [^\s]+#0 : [^\s]+\)
+\([^\s]+guard[^\s]+ \? [^\s]+ : [^\s]+#0\)
 --
 These regexes are making sure that a variable of generation 0 dosen't appear in a phi merge, so the below
 statement:

regression/cbmc/phi-merge_uninitialized_values/static_local.desc

@@ -5,8 +5,8 @@ activate-multi-line-match
 ^EXIT=0$
 ^SIGNAL=0$
 --
-^.*:\s+staticLocal(@[0-9]+)?#0\)
-^.*\?\s+staticLocal(@[0-9]+)?#0\s+:
+\([^\s]+guard[^\s]+ \? [^\s]+#0 : [^\s]+\)
+\([^\s]+guard[^\s]+ \? [^\s]+ : [^\s]+#0\)
 --
 These regexes are making sure that a variable of generation 0 dosen't appear in a phi merge, so the below
 statement: