virtualization: add new builtin command to print hydration level (microsoft#659)

GVFS users can easily (and accidentally) over-hydrate their enlistments.
This causes some commands to be very slow.

Create a command to print the current hydration level. This should help
our support team investigate the state of their enlistment.

This command will print something like:

```
% git virtualization
Skipped: 2
Hydrated: 3
Total: 5
Hydration: 60.00%
```

and log those values to Trace2 in a `data_json` record of the form:

```
{"skipped":2,"hydrated":3,"total":5,"hydration":60.00}
```

Author: dscho

.github/workflows/scalar-functional-tests.yml

@@ -20,7 +20,7 @@ jobs:
       fail-fast: false
       matrix:
         # Order by runtime (in descending order)
-        os: [windows-2019, macos-11, ubuntu-20.04, ubuntu-22.04]
+        os: [windows-2019, macos-13, ubuntu-20.04, ubuntu-22.04]
         # Scalar.NET used to be tested using `features: [false, experimental]`
         # But currently, Scalar/C ignores `feature.scalar` altogether, so let's
         # save some electrons and run only one of them...

t/t1093-virtualfilesystem.sh

@@ -69,6 +69,8 @@ test_expect_success 'verify status is clean' '
 	git status > actual &&
 	cat > expected <<-\EOF &&
 		On branch main
+		You are in a partially-hydrated checkout with 75% of tracked files present.
+
 		nothing to commit, working tree clean
 	EOF
 	test_cmp expected actual

wt-status.c

@@ -1604,10 +1604,15 @@ static void show_sparse_checkout_in_use(struct wt_status *s,
 {
 	if (s->state.sparse_checkout_percentage == SPARSE_CHECKOUT_DISABLED)
 		return;
-	if (core_virtualfilesystem)
-		return;
-
-	if (s->state.sparse_checkout_percentage == SPARSE_CHECKOUT_SPARSE_INDEX)
+	if (core_virtualfilesystem) {
+		if (s->state.sparse_checkout_percentage == SPARSE_CHECKOUT_SPARSE_INDEX)
+			status_printf_ln(s, color,
+					 _("You are in a partially-hydrated checkout with a sparse index."));
+		else
+			status_printf_ln(s, color,
+					 _("You are in a partially-hydrated checkout with %d%% of tracked files present."),
+					 s->state.sparse_checkout_percentage);
+	} else if (s->state.sparse_checkout_percentage == SPARSE_CHECKOUT_SPARSE_INDEX)
 		status_printf_ln(s, color, _("You are in a sparse checkout."));
 	else
 		status_printf_ln(s, color,
@@ -2560,6 +2565,36 @@ void wt_status_print(struct wt_status *s)
 			   s->untracked.nr);
 	trace2_data_intmax("status", s->repo, "count/ignored", s->ignored.nr);
 
+	switch (s->state.sparse_checkout_percentage) {
+	case SPARSE_CHECKOUT_DISABLED:
+		break;
+	case SPARSE_CHECKOUT_SPARSE_INDEX:
+		/*
+		 * Log just the observed size of the sparse-index.
+		 *
+		 * When sparse-index is enabled we can have
+		 * sparse-directory entries in addition to individual
+		 * sparse-file entries, so we don't know the complete
+		 * size of the index.  And we do not want to force
+		 * expand it just to emit some telemetry data.  So we
+		 * cannot report a percentage for the space savings.
+		 *
+		 * It is possible that if the telemetry data is
+		 * aggregated, someone will have a good estimate for
+		 * the size of a fully populated index and can compute
+		 * a percentage after the fact.
+		 */
+		trace2_data_intmax("status", s->repo,
+				   "sparse-index/size",
+				   s->repo->index->cache_nr);
+		break;
+	default:
+		trace2_data_intmax("status", s->repo,
+				   "sparse-checkout/percentage",
+				   s->state.sparse_checkout_percentage);
+		break;
+	}
+
 	trace2_region_enter("status", "print", s->repo);
 
 	switch (s->status_format) {