Higher RAM usage in NETCore

### Is there an existing issue for this?

- [X] I have searched both open/closed issues, no issue already exists.

### CefSharp Version

120.2.50

### Operating System

Windows 11

### Architecture

x86

### .Net Version

.NET 6.0

### Implementation

WinForms

### Reproduction Steps

Create minimal application that uses CefSharp and build it both for .NET Framework and .NET Core.

### Expected behavior

Both versions should take a similar amount of RAM.

### Actual behavior

There's a 20-30% difference between the two versions' working set.

See this table (taken from Process Explorer):

.NET 6:
| Process name | Private Bytes | Working Set |
|--------|--------|--------|
| Chromium.exe | 37,292 K | 99,908 K |
| CefSharp.BrowserSubprocess.exe | 101,244 K | 85,808 K |
| CefSharp.BrowserSubprocess.exe | 21,888 K | 34,820 K |
| CefSharp.BrowserSubprocess.exe | 23,264 K | 42,092 K |
| CefSharp.BrowserSubprocess.exe | 25,764 K | 43,848 K |
| CefSharp.BrowserSubprocess.exe | 28,040 K | 58,440 K |

.NET 4.7.2:
| Process name | Private Bytes | Working Set |
|--------|--------|--------|
| Chromium.exe | 35,532 K | 80,800 K |
| CefSharp.BrowserSubprocess.exe | 103,972 K | 75,220 K |
| CefSharp.BrowserSubprocess.exe | 21,380 K | 32,504 K |
| CefSharp.BrowserSubprocess.exe | 20,404 K | 25,964 K |
| CefSharp.BrowserSubprocess.exe | 24,264 K | 34,704 K |
| CefSharp.BrowserSubprocess.exe | 26,220 K | 48,696 K |

This increases even further when I actually start browsing in either program.

### Regression?

_No response_

### Known Workarounds

_No response_

### Does this problem also occur in the CEF Sample Application

Not Tested

### Other information

_No response_

Start by reproducing the reported minimal CefSharp WinForms application on Windows 11 x86 with .NET 6.0 and .NET Framework 4.7.2. Use Process Explorer to compare Chromium.exe and CefSharp.BrowserSubprocess.exe private bytes and working sets before and during browsing. Done means identifying the cause of the 20–30% difference and bringing the runtimes' memory usage to a similar level, or documenting a confirmed runtime limitation.

Written by the indexing model from the issue text.