Improve readibility of compare_to_baseline output message (#2025)

Summary:

Integrating the following feedback into the construction of the compare_to_baseline output message

 " I'm noticing this could be a bit easier to digest. Luckily we can use markdown formatting as follows:
* add an asterisk before each item in comparison_list so that the user gets a bulleted list.
* put metric name + percent improvement first, followed by arm names + absolute metric values, since the former tends to be what the user most cares about.
* bolding the first part of each line item can help create visual anchors for the reader.
* adding backticks can help users identify variable names/values. "

Reviewed By: bernardbeckerman, mpolson64

Differential Revision: D51632256

ax/service/tests/test_report_utils.py

@@ -1013,8 +1013,8 @@ def test_compare_to_baseline_moo(self) -> None:
             comparison_arm_names = ["opt_0", "opt_1_min", "opt_3"]
 
             preamble = (
-                "Each of the following arms optimizes a different "
-                "objective metric.<br>"
+                "Below is the greatest improvement, if any,"
+                " achieved for each objective metric \n"
             )
             output_text_0 = _format_comparison_string(
                 comparison_arm_name="opt_0",
@@ -1054,10 +1054,11 @@ def test_compare_to_baseline_moo(self) -> None:
 
             expected_result = (
                 preamble
+                + " * "
                 + output_text_0
-                + "<br>"
+                + " * "
                 + output_text_1
-                + "<br>"
+                + " * "
                 + output_text_3
             )
 

ax/service/utils/report_utils.py

@@ -1148,12 +1148,10 @@ def _format_comparison_string(
     digits: int,
 ) -> str:
     return (
-        f"{comparison_arm_name=} "
-        + "improves your objective metric "
-        + f"{objective_name} by {percent_change:.2f}%. "
-        + f" {baseline_arm_name=} was improved "
-        + f"from {baseline_value=:.{digits}f}"
-        + f" to {comparison_value=:.{digits}f}"
+        "**Metric "
+        f"`{objective_name}` improved {percent_change:.{digits}f}%** "
+        f"from `{baseline_value:.{digits}f}` in arm `'{baseline_arm_name}'` "
+        f"to `{comparison_value:.{digits}f}` in arm `'{comparison_arm_name}'`.\n "
     )
 
 
@@ -1339,17 +1337,17 @@ def compare_to_baseline(
     result_message = ""
     if len(comparison_list) > 1:
         result_message = (
-            "Each of the following arms optimizes a different "
-            + "objective metric.<br>"
+            "Below is the greatest improvement, if any,"
+            " achieved for each objective metric \n"
         )
 
-    for idx, result_tuple in enumerate(comparison_list):
+    for _, result_tuple in enumerate(comparison_list):
         comparison_message = _construct_comparison_message(*result_tuple)
         if comparison_message:
             result_message = (
                 result_message
+                + (" * " if len(comparison_list) > 1 else "")
                 + not_none(comparison_message)
-                + ("<br>" if idx != len(comparison_list) - 1 else "")
             )
 
     return result_message if result_message else None