[Impeller] ensure precision is matched between fragment and vertex stage for PowerVR GPU bug

impeller/entity/shaders/blending/porter_duff_blend.vert

@@ -16,7 +16,7 @@ in vec2 texture_coords;
 in vec4 color;
 
 out vec2 v_texture_coords;
-out f16vec4 v_color;
+out mediump f16vec4 v_color;
 
 void main() {
   gl_Position = frame_info.mvp * vec4(vertices, 0.0, 1.0);

impeller/entity/shaders/glyph_atlas.vert

@@ -25,7 +25,7 @@ in vec2 glyph_position;
 
 out vec2 v_uv;
 
-IMPELLER_MAYBE_FLAT out f16vec4 v_text_color;
+IMPELLER_MAYBE_FLAT out mediump f16vec4 v_text_color;
 
 mat4 basis(mat4 m) {
   return mat4(m[0][0], m[0][1], m[0][2], 0.0,  //

impeller/entity/shaders/position_color.vert

@@ -13,7 +13,7 @@ frame_info;
 in vec2 position;
 in vec4 color;
 
-out f16vec4 v_color;
+out mediump f16vec4 v_color;
 
 void main() {
   gl_Position = frame_info.mvp * vec4(position, 0.0, 1.0);

impeller/entity/shaders/solid_fill.vert

@@ -12,7 +12,7 @@ frame_info;
 
 in vec2 position;
 
-IMPELLER_MAYBE_FLAT out f16vec4 v_color;
+IMPELLER_MAYBE_FLAT out mediump f16vec4 v_color;
 
 void main() {
   v_color = frame_info.color;

impeller/entity/shaders/texture_fill.vert

@@ -16,7 +16,7 @@ in vec2 position;
 in vec2 texture_coords;
 
 out vec2 v_texture_coords;
-IMPELLER_MAYBE_FLAT out float16_t v_alpha;
+IMPELLER_MAYBE_FLAT out mediump float16_t v_alpha;
 
 void main() {
   gl_Position = frame_info.mvp * vec4(position, 0.0, 1.0);