Use proper RTS flags when collecting residency in perf tests.

Replace options like collect_stats(['peak_megabytes_allocated'],4) with
collect_runtime_residency(4) and so forth. Reason being that the later
also supplies some default RTS arguments which make sure residency does
not fluctuate too much.

The new flags mean we get new (hopefully more accurate) baselines so
accept the stat changes.

 -------------------------
 Metric Decrease:
     T4029
     T4334
     T7850
 -------------------------

testsuite/tests/perf/should_run/all.T

@@ -88,15 +88,15 @@ test('T4321',
 test('T3736', [], makefile_test, ['T3736'])
 test('T3738',
      [extra_clean(['T3738a.hi', 'T3738a.o']),
-      collect_stats('peak_megabytes_allocated', 0),
+      collect_runtime_residency(1),
       collect_stats('bytes allocated',8),
-      only_ways(['normal'])
+      only_ways(['normal']),
       ],
      compile_and_run,
      ['-O'])
 
 test('MethSharing',
-     [collect_stats('peak_megabytes_allocated',  0),
+     [collect_runtime_residency(1),
       collect_stats('bytes allocated',5),
       only_ways(['normal'])
       ],
@@ -177,7 +177,7 @@ test('T5536',
 
 test('T7257',
      [collect_stats('bytes allocated',10),
-      collect_stats('peak_megabytes_allocated',5),
+      collect_runtime_residency(5),
       only_ways(['normal'])
      ],
     compile_and_run, ['-O'])
@@ -192,14 +192,14 @@ test('T7507', omit_ways(['ghci']), compile_and_run, ['-O'])
 # For 7507, stack overflow is the bad case
 
 test('T7436',
-     [collect_stats('max_bytes_used',4),
+     [collect_runtime_residency(4),
       only_ways(['normal'])
       ],
      compile_and_run,
      ['-O'])
 
 test('T7436b',
-     [collect_stats('max_bytes_used',4),
+     [collect_runtime_residency(4),
       only_ways(['normal'])
       ],
      compile_and_run,
@@ -221,7 +221,7 @@ test('T7954',
      ['-O'])
 
 test('T7850',
-     [collect_stats('peak_megabytes_allocated',10),
+     [collect_runtime_residency(10),
       only_ways(['normal'])],
      compile_and_run,
      ['-O'])
@@ -308,7 +308,7 @@ test('T12990',
 
 test('T13218',
     [collect_stats('bytes allocated',5),
-     collect_stats('max_bytes_used',10),
+     collect_runtime_residency(10),
      only_ways(['normal'])],
     compile_and_run,
     ['-O'])

testsuite/tests/perf/space_leaks/all.T

@@ -3,7 +3,7 @@ test('space_leak_001',
      # This could potentially be replaced with
      # collect_stats('all',5) to test all 3 with
      # 5% possible deviation.
-     [collect_stats(['peak_megabytes_allocated','bytes allocated'],5),
+     [ collect_stats(['bytes allocated'],5),
       collect_runtime_residency(15),
       omit_ways(['profasm','profthreaded','threaded1','threaded2','nonmoving_thr'])
       ],
@@ -13,7 +13,7 @@ test('space_leak_001',
 test('T4334',
      # Test for a space leak in Data.List.lines (fixed with #4334)
      [extra_run_opts('1000000 2 t'),
-      collect_stats('peak_megabytes_allocated',2),
+      collect_runtime_residency(2),
       # prof ways don't work well with +RTS -V0, nonmoving way residency is
       # highly variable.
       omit_ways(['profasm','profthreaded','nonmoving_thr'])
@@ -21,9 +21,8 @@ test('T4334',
      compile_and_run, [''])
 
 test('T2762',
-     [collect_stats('peak_megabytes_allocated',2),
+     [collect_runtime_residency(2),
       only_ways(['normal']),
-      extra_run_opts('+RTS -G1 -RTS' ),
       extra_clean(['T2762A.hi', 'T2762A.o'])],
      compile_and_run, ['-O'])
 
@@ -32,9 +31,7 @@ test('T4018',
      compile_and_run, ['-fno-state-hack'])
 
 test('T4029',
-     [collect_stats(['peak_megabytes_allocated'],10),
-      collect_runtime_residency(10),
-      extra_hc_opts('+RTS -G1 -RTS' ),
+     [collect_runtime_residency(10),
       ],
      ghci_script,
      ['T4029.script'])